<insert figure 6.1 here> Figure 6.1. Distribution of sites with fifth millennium radiocarbon dates (1. Canal Rd. Muirtown, 2. Tarradale, The Black Isle, 3. Skilmafilly, 4. Milltimber, 5. Nethermills, 6. Spurryhillock, 7. Risga, 8. Carding Mill Bay, 9-12. Caisteal nan Gillean I and II Cnoc Coig, Priory Midden, Cnoc Sligeach, 13. Morton B, 14. Airthrey, 15. Blair Drummond, 16. Braehead, 17. Causewayhead, 18. Meiklewood, 19. Inveravon, 20. Mumrills, 21. Polmonthill, 22. Kilrubie Hill, 23. Wide Hope Shank, 24. Flint Hill, 25. Burnetland Hill, 26. Low Hauxley, 27. Stainton West, 28. South Haw, 29. Slynedales Culvert, 30. Rocher Moss, 31. March Hil Top, 32. March Hill Carr, 33. Lominot Site 2, 3 and C, 34. Dan Clough, 35. Parc Bryn Cegin, 36. Nant Hall, Prestatyn, 37. Dunford Bridge B, 38. Rough Close, 39. Lydstep, 40. Goldcliff, 41. Ascott-under-Wychwood, 42. Manor Farm, 43. Stratford's Yard, 44. Misbourne Viaduct, 45. Tolpitts lane, 46-8, Wawcott I, III, XXIII, 49. Langley’s Lane, 50. Blick Mead, 51. Coneybury Anomaly, 52. Fir Tree Field Shaft, 53. Blashenwell, 54 . Little Dartmouth Farm, 55. Poldowrian, 56. Windmill Farm, 57. Charlwood, 58. Bexhill, 59. Falmer Stadium). Chapter 6 The Last Hunters: The Final Mesolithic 5000-4000BC ….the mark they could make on the face of Britain must have been slight indeed. A few trees cut - extending here and there to a small clearing; boats moving on the rivers and along the shore… J. Hawkes 1951, 153-4 The Final Mesolithic in Britain is a story of both continuity and change. Many favoured locations that had been used for millennia remained important. At the same time in some areas new ways of engaging with the landscape appear. On the small island of Oronsay and the estuaries of the east coast of Scotland people accumulated shell to produce middens on a monumental scale; on Oronsay these are associated with the deposition of human remains. In the Pennines there appears a greater intensity of occupation in the fifth millennium, at least as far as radiocarbon dated sites indicate. In this region in the final centuries of the period there were further changes, as new microlith types emerge and new material sources were used; these may indicate shifting mobility patterns or the emergence of new social contacts. New forms of material culture appeared, often referencing distant places. T-axes, the type fossil of the southern Scandinavian Ertebølle, are found in Scotland (Elliott 2018). Polished axes in Cumbria may suggest contact across the Irish sea (Brown in press). New microlith forms are found across southern Britain, south Wales, and as far north 1 as Cumbria and County Durham (figure 6.2). Amongst these are micro-tranchet microliths, which are similar in concept to the trapezes and transverse arrowheads of late Mesolithic continental Europe but rendered through a lens of traditional British late Mesolithic microlithisation, <insert figure 6.2 here> Figure 6.2 Distribution of sites with likely fifth millennium microlith types: four-sided microliths (left); asymmetric microtranchets (centre) and symmetrical micro-tranchets (right) It may be pertinent to ask to what extent these changes in Britain were stimulated by developments on the continent. By c.5000 cal BC Neolithic groups had reached the northern French coast. Populations and ideas were shifting. Long distance east-west maritime voyaging has recently been identified in the Channel (Anderson-Whymark et al 2015, Conneller et al. 2016); it seems unlikely that north-south travel did not also occur. The bones of domestic cattle dating to 4495–4195 cal BC (Woodman & McCarthy 2003, 33) have been found at the Mesolithic site of Ferriter’s Cove on the southwest coast of Ireland, suggested to represent gifts of meat from Neolithic groups far to the south in continental Europe. While no domesticated animals have thus far been identified in Britain, claims have been made for the small-scale cultivation of cereals in the last centuries of the fifth millennium (Blackford and Innes 2020). The changes that can be seen in Britain at this time may suggest new fusions of cultures and ideas. Sea-level rise and vegetation change in the fifth millennium By 5000 cal BC Britain was approaching its current form (figure 6.3). A handful of small islands in the Doggerland archipelago perhaps remained. The shoreline of southern Britain extended further than today, particularly in the region of the Wash where it is likely to have been tens of kilometres further east (Walker et al. 2020). The end of the period is marked by a climatic downturn, the 5.9ka event. The effects of this are seen in European lake sequences from about 4000 cal BC (Florescu 2019). Tipping (2010b) argues that lower sea temperatures and increased storminess between 4500 and 3500 cal BC may have altered movements of migratory fish and changed the form of coastal landscapes through erosion and increased dune formation. More storms may have increased the risks involved in maritime travel (ibid.), though their effects are likely to have been more pronounced in the North Sea in comparison to the western seaways (Garrow and Sturt 2011). <insert figure 6.3 here> Figure 6.3. Palaeogeographic model of Britain at 5000 cal BC (Copyright Fraser Sturt) While all major tree species were present in Britain by 5000 cal BC, the slower-spreading late arrivals, ash and lime continued to expand, and other species moved towards their natural limits. As sea-levels rose, rivers backed up, leading to increasing development of wetlands and the boggy ground favoured by alder. Peat seems to have increased in the uplands and here too human activities seem to have had an impact, with burning used to create clearings, favour certain species such as hazel and even create open heathland in the Pennines and North York Moors. 2 Northeast Britain The settlement pattern in the late Mesolithic was characterised by sites on rivers and these are likely to have remained important in the Final Mesolithic. A late date of 4325-4055 cal BC (Poz-69104) may indicate continued Final Mesolithic occupation at Nethermills, Deeside (Wickham-Jones et al. 2016; see chapter 5); this is on oak though so could conceivably fall into the early Neolithic due to the old wood effect. Estuarine sites seem particularly important in this period, with shell-middens dating mainly to the last millennium of the Mesolithic preserved on raised shorelines along the western edge of the Firth of Forth in particular. Coastal occupation is also in evidence, consisting of small-scale but repeated visits to the shore and the generation of small shell-middens at Morton B (1971). Occupation also continued in the uplands: In the Cairngorms, the Chest of Dee, an important halting point in the seventh millennium, also has fifth millennium dates (Wickham-Jones 2020). Further south in the Southern Uplands Biggar Archaeology Group’s site at Weston Farm also has late dates (Ward 2017). Daer 3 and 84 date to the end of the fifth millennium BC as does the quarry site at Burnetland (see below). At Garvald Burn, at 375m asl, a series of pits has three dates in the fifth millennium BC (Ballin and Barrowman 2015). Further to the west, occupation continued around Loch Doon at Smittons T3 in the late fifth millennium, close to the late sixth millennium site of Smitton T1 (Finlayson 1990). Continuity can perhaps best be seen at Milltimber, where pits had been dug for millennia. This site also has fifth millennium dates; similar ways of engaging with this place seem to have stretched across millennia. Post-holes associated with pits were also in evidence in the central areas of the site; the only radiocarbon measurement from these post-holes is between 4895-4710 cal BC (SUERC-39748) (Dingwall et al. 2019). This focus on erecting postholes at a time of rapid sea-level change may indicate an increased concern with marking actual physical places, whereabouts of which might become disguised in the face of moving ecotonal zone. The digging of large pits also commenced at new sites without these long histories. On the flanks of a small hill 30km north of Aberdeen, at Skilmafilly, a large pit, measuring 3.2m by 2.8m and 1.4m deep was excavated (Johnson and Cameron 2012). As with the large Middle Mesolithic pits in this region, it was left to silt up naturally, before twice being recut. Both the initial digging of this pit and the recuts were undertaken in the late 5th millennium BC. At Spurryhillock, 20km south of Aberdeen, a large truncated pit measuring 2.3 by 1.8m, and surviving to a depth of 1.3m, silted up with alternate layers of charcoal and sand (Alexander 1997). Oak charcoal from the lowest fill indicates it was dug in the early-mid fifth millennium BC, but the old wood effect means it may be broadly contemporary with Skilmafilly. Textbox 6.1. Mesolithic quarrying In previous chapters, several locations have been identified as places for the procurement of lithic raw material. In southern England, Mesolithic groups have been argued to exploit solution hollows and tree throws as means to access flint nodules (Care 1982). Broom Hill, located close to the chalk, where known solution pipes exist is such a site, and the large lithic assemblage containing a substantial quantity of axes supports this interpretation. The material recovered from two solution pipes near Fort Wallingford during construction of the M27, and also associated with large numbers of axes and cores, could be interpreted in a similar way (Hughes and ApSimon 1977). A case has also been made that some of the so- called pit houses, excavated in the first half of the 20th century may also have been quarry pits (see chapter 4). Higgs (1959) has suggested the pit at Downton was cut to exploit gravel flint, and similar arguments could be made that some of the Farnham pits were used to procure flint. More recently it has been suggested that amorphous pits at Streat Lane, 3 Sussex (Butler 2007), London Road, Beddington (Bagwell et al. 2001) and Woodbridge Road, Guildford (Bishop 2008) may have been dug to quarry flint from Head and terrace gravel deposits respectively. At both London Road and Woodbridge Road, it has been suggested that tree throws may have alerted people to the presence of flint in the underlying gravels. While earlier procurement sites are focused mainly on secondary flint sources, in the Final Mesolithic quarry sites associated with the exploitation of primary chert deposits seem to appear. Survey by the Biggar Archaeology Group located at least eight quarry pits and associated spoil heaps on the southwest flank of Burnetland Hill in the Scottish Borders, a known source of Southern Uplands chert. The largest of the depressions were around 5m across with spoil heaps 1.5m high. Excavations of one of these revealed it was excavated against the edge of a number of seams of radiolarian chert, of the blue-grey sort favoured in prehistory. A number of large stone pounders were found, used to remove material from the seams. Knapped chert, including bladelet production, was recovered from the quarry pits. Measurements on hazel charcoal recovered from the base of the trench suggested a very late Mesolithic date for quarrying of 4225-3960 cal BC (5220+-35BP; SUERC-17876) (Ward 2012). Quarry pits of similar size are known from three further sites in the Upper Tweed Valley, located through surveys undertaken by Bob Knox at Flint Hill, Kilrubie Hill and Wide Hope Shank (Warren 2001). Additional survey by Warren recorded around 57 quarry pits across the three sites, with a range of 2-10m in size. Excavations of one of the pits at Wide Hope Shank (figure 6.4) revealed a very similar type of archaeology to Burnetland, with large quantities of shattered chert and similar, though fragmentary, hammerstones recovered (ibid.). People appear to have been cutting back the poor quality, frost fractured surface exposures to reach better quality material. Some burnt chert might indicate the use of fire in quarrying but no charcoal was found. These three quarries remain undated, and only undiagnostic material was associated with the spoil heaps, though blade debitage was found close by. The dating of Burnetland reinforces suspicions that some of these quarry pits are of Mesolithic date. Warren (2001, 218) points to the presence of chert with surfaces indicative of procurement direct from an exposure at several Tweed Valley sites including, Manor Bridge, Peebles, dated to the second half of the ninth millennium BC, indicating this practice may have some antiquity. <insert figure 6.4 here> Figure 6.4. Quarry pits at Wide Hope Shank (Copyright Graeme Warren, UCD School of Archaeology) In the late Mesolithic the chert sources of northern and southwest England were increasingly used, but evidence for quarrying is more ambiguous. Jeff Radley notes ‘chert diggings in Kirkdale below Seldon’, suggesting that he may have located evidence of quarrying, though this is undated (Radley archive, notebook 2, cited in Myers 1986, 368). Survey by Hinds (1998) have failed to locate any such evidence. There have also been suggestions that Portland chert was quarried (Care 1979, 99), but there is currently no evidence remaining to support this, though this may have been destroyed by later quarries. There have also been claims for Mesolithic quarrying of meta-mudstone at Carn Menyn (Darville and Wainwright 2015). Four Mesolithic-age radiocarbon dates have been obtained from the primary fills of a quarry pit. These suggest two phases of activity: one in the first half of the seventh millennium BC, and a second at the end of the sixth. There is as yet though 4 no evidence for the use of meta-mudstone on Mesolithic sites in Wales (David pers comm.), raising the possibility that the charred materials are residual. Eastern Middens Shell middens have in general been associated with Mesolithic lifeways on the west coast, but they are present in the east too and not simply at the better known but rather atypical middens at Morton B (see below). Most other middens were estuarine, occupied at a time of much higher sea-level, and now far inland. Several cluster around the inner Forth estuary and tend, unlike Morton, to comprise mainly of oyster shell. Others are found overlooking the Beauly and Moray Forth. Most of these are early finds that are poorly dated and few have undergone systematic excavation. A tendency to be located on raised shorelines has led to the suggestion that they are Mesolithic; however several have Neolithic evidence in the form of pottery or radiocarbon dates, suggesting a more complex situation and it is likely that many accumulated during both the Mesolithic and the Neolithic, as is the situation on the west coast. Several of these large estuarine middens do have Mesolithic dates, and while some earlier examples are known, most do seem to date to the fifth millennium BC, though this may be an issue of preservation, rather than changing practices. A recent survey and excavation programme by the North of Scotland Archaeological Society has located eight middens at Tarradale on The Black Isle, overlooking the northern shore of the Beauly Firth (Grant 2018). A midden (site 2D) on the 17m asl raised beach dates to between 6630 and 6000BC (Waddington and Wicks 2017), the earliest recorded in the east. It was accumulated at a time of high sea level, slightly before the maximum postglacial transgression. A second midden (2B) has a series of radiocarbon dates (4782-3643 cal BC) spanning most of the fifth millennium and into the Neolithic (Grant 2018) and is located on the 9m asl raised beach, a few metres above a sea that was beginning to retreat. Found within this midden were a biserial antler point and two antler T-axes (textbox 6.4, figure 6.5), both artefacts also associated with middens in the west of Scotland. Stone settings were also located, and possible structural evidence. This may indicate a similar pattern to Cnoc Coig, with midden deposits building up around a dwelling structure. The midden has a range of shellfish (oyster, mussels, cockles and periwinkles), fish, birds and terrestrial fauna, though the range of species reported, including domesticates, indicates that it is of mixed date. A third midden (2A) similarly spans the centuries of the Mesolithic-Neolithic transition and indicates continuity of occupation and the form of activities across this period (Grant 2018). <insert figure 6.5 here> Figure 6.5. T-axe from midden 2B at Tarradle, on The Black Isle. Copyright Tarradale Through Time; photo by Michael Sharpe On the southern shore of the Beauly Firth, a date of 4650-4345 cal BC (GU-1473) comes from the basal layers of a midden at Canal Rd, Muirtown, one of three known from Inverness (Myers and Gourlay 1991), though the measurements were made on possibly mature oak timbers, and could thus belong to a Neolithic occupation. This midden, no more than 0.6m thick, extended intermittently for around 100m along the ancient shoreline. The midden was mainly composed of oyster, though mussels, cockles and winkles were found in small concentrations and fragments of crab shell were noted. While a hearth was present, no lithic material was recovered. 5 In the Forth Valley at the time of the main post-glacial transgression at around 5000-5500 cal BC relative sea-level was at c12m, with mean spring tide level at c14-16m, turning the Forth Valley into a large marine embayment with estuarine conditions at the western end (Smith et al. 2010, Bradley et al. 2011). Areas of mudflats and salt marshes increased as relative sea- level fell (Smith et al. 2010). In the western end of the Firth of Forth Mesolithic dates have been obtained from a midden at Inveravon, on the southern shore of the estuary, now on the north bank of the river Avon. The midden stood nearly 3m high and stretched for 120m. It was composed mainly of oyster shell, with some bones observed (Deveraux and Sloane 1983). Five measurements, one on charcoal and four on shell, suggest it began to accumulate during the sixth and fifth millennium cal BC, with the upper layers being Neolithic (Mackie 1972, Sloane 1986). The midden had the remains of hearths set amongst the shells, but no artefacts were found. Only small sections of this midden were investigated when it was disturbed by a pipe trench in 1971 and 1983. A second midden in the same area, Mumrills, just 3km from Iveravon, has radiocarbon dates on shell from the base of the midden belonging to the late fifth millennium BC (Ashmore 2004); though the upper layers here too were Neolithic. Also close to Inveravon, on the south bank of the Avon, a third, undated midden at Polmonthill was reported to have been 150m long, 32m wide and around a metre in height. Both Polmonthill and Inveravon were composed mainly of oyster with occasional mussel, cockle, periwinkle and whelk (Lacaille 1954). All three middens were located on the 15m contour, around three metres above the sea level of the time of the maximum transgression. Around 15km to the northwest at the most westerly point of the estuary at the time of occupation, a fourth small midden at Braehead, composed mainly of oysters, but with scallops, winkles and mussels, was dated to the second half of the fifth millennium cal BC (Ashmore and Hall 1996). The Whale hunters In the nineteenth century, several finds of whale skeletons associated with artefacts of red deer were made around Stirling. Clark (1947) associated these with deposits called carse clays: intertidal and sub-tidal deposits from a time of higher relative sea-level. He suggested that these were animals that had swum up the Firth of Forth and become beached, before being exploited by human groups. The best known, and the only dated example, comes from Meiklewood. This whale was associated with an antler T-axe (Elliott 2012, textbox 6.4), which has been dated to 5000-4590 cal BC (5920±80BP; OxA-1159) (Bonsall and Smith 1989, 36). This was found ‘resting upon the front of the [whale] skull, lying vertically in the blue silt’ (Turner 1889, 791). However a recent date on a vertebra of the associated whale (a rorqual) at 7590-7190 cal BC (8400±80BP; Beta-158485) (Smith et al. 2010) is much earlier than the date of the mattock. Given the broader pattern of whales associated with antler tools, this association is unlikely to be a coincidence, suggesting one of these measurements is wrong, probably affected by either organic or inorganic conservation media. Dates for antler mattocks across Europe suggest the measurement on the whale bone is erroneous. The other whales and associated artefacts are undated, though their stratigraphic position, in the carse clays but usually resting on the underlying peat (Smith et al. 2010) suggests they are also Mesolithic in date. The only other surviving antler object, from Causewayhead, appears a more informal tool, consisting of part of the beam and an attached tine (Lacaille 1954). However the original descriptions of the finds suggest that the now-lost antler tools recorded may have been T-shaped antler axes. At Blair Drummond ‘a piece of perforated deer’s antler with traces of a wooden handle’ was found, while a hole had been bored in one of the two antlers found at Airthrey (Turner, 1889, 790). Turner (ibid.) suggests these implements would have been used as blubber mattocks. These whales have rarely been identified to species. The Airthrey animal is reported to have been a 22-metre-long blue whale, while the find from Meiklewood was a species of rorqual 6 (Clark 1947). The most common modern strandings on British coastlines are porpoises and dolphins, with an average of 540 reported per year over the past 5 years; most of these as a result of injuries from nets and trawlers (Deaville 2016). Whale strandings are rarer (table 6.1). Sperm whales average at around six per year, and are well-represented in Scotland, particularly in the northern and western islands. Many animals are washed up dead but live strandings, both singly and en masse are not uncommon, and there have been suggestions that the North Sea functioned as a sperm whale trap (Scottish Agricultural College 2000). The quantity of resources that could be obtained from such a beaching event cannot be over-estimated. Males can be up to 20.5m long and were hunted in historic times for spermaceti and blubber (used in oil lamps and to make candles) and the flesh was sometimes eaten. A large whale could provide as much as 500 gallons of spermaceti. Olaus Magnus in the sixteenth century estimated that a single whale could provide 250-300 waggons full of usable products, with meat for salting, blubber for heating, small bones for fuel, and large bones for building (cited in Clark 1947, 90) Clark (1947) notes that the majority of prehistoric whale finds are rorquals. This group consist of several species, of which five have been washed up on British beaches. The largest of these is the blue whale which can be up to 30m in length, while even the smallest, the Minke whale, is up to 9m in length. Apart from Minke whales with an average of 15 strandings per year, these are rarely found on British beaches. Most Minkes died from entanglement in commercial fishing gear and this was also a factor in the death of several of the rarer whales. Whale populations have been decimated by hunting, driving several species to near extinction; numbers are only beginning to recover. Even so, a stranding in the Mesolithic would probably have been a relatively rare event; blue whales in particular a once in a lifetime encounter with a perhaps fabled species. Rorqual species Average strandings per year 2011-15 Minke whale (Balaenoptera acutorostrata) 15 Fin whale (Balaenoptera physalus) 3 Humpback whale (Megaptera novaeangliae) 1 Sei Whale (Balaenoptera borealis) 1 Blue whale (Balaenoptera musculus) 0 (last reported stranding in 1957) Table 6.1. Strandings of rorquals 2011-2015 on UK beaches (data from Deaville 2016). The whales stranded on inter-tidal mud flats, the middens along the shoreline behind, offer a vivid picture of life in this estuarine inlet in the fifth millennium BC. Around a quarter of finds of whales have associated artefacts, and other examples may have been over looked during early discoveries, indicating Mesolithic people were routinely present in these areas and whale sightings may have been monitored. The excess in resources from a stranding could have been the occasion for feasting and gifting, creating ties of obligation between local groups. Textbox 6.2. Life on the Coast: Morton B At Morton B, on the eastern side of the same former small rocky island as the earlier site of Morton A (see chapters 2 and 5), excavations uncovered a shell-midden around 30m long, 3.5m wide and up to 70cm deep in places, but more typically between 10cm and 45cm in depth. Dates on bone artefacts from the midden probably suggest intermittent occupation on a number of different occasions during the fifth millennium BC, which would have been just after the time of the maximum transgression, when sea-levels would have been around 7-8m asl (Bradley et al. ibid). A raised beach at 9.7m asl has been recorded, and the midden built 7 up on top of beach deposits of cobbles and shingle. Land snails from the base of the midden indicate that a heavily shaded, dry woodland would have been present nearby (Coles 1983). The midden was composed of shell, stones, fish, birds and animal bone. Faunal remains were highly fragmentary and only a small percentage could be identified to species, but red deer, roe deer, aurochs, pig and hedgehog were all recovered. Birds were common, particularly guillemot and gannet, but include fulmar, cormorant, shag, razorbill, puffin, back- backed gull, kittiwake, thrush and crow. Many of these favour cliff ledges and suitable rock formations (also a possible source of the lava used on site) can be found around 1.5km from the site. Of fish remains cod were most common, but haddock, turbot, sturgeon, salmon or sea-trout also represented. The presence of cod has been suggested to be indicative of deep-water fishing (Coles 1971, 353, but see Pickard and Bonsall 2004). Only a small sample of the shell material was examined, but this revealed 40 different species were present. Most were marine shells of which cockle, Baltic clam and striped venus clam were most common and limpet, common and flat periwinkle, dogwelk, common welk, ocean quahog well-represented, but brackish water species were also present indicating use of estuarine resources. Plant remains were present in small numbers; all species known to have been eaten historically, such as fat hen, iron root, knotgrass, corn spurrey and chickweed. A notable absentee from the middens is seal (Coles 1983, 12), particularly given the probable presence of rocky islands in the area. Seal bones are present at some of the Middle Mesolithic sites in the region and argued to have been an important component of late Mesolithic diets in the region (Waddington 2007). Distinct dumping episodes of shells could be discerned in places as could periods of abandonment, when the surface of the midden began to erode. Localised deposits of flat stones and boulders were noted, including an area where the midden appeared to have been levelled then covered with stones. Also within the midden were stone-set hearths, posts and stakeholes. None of these were arranged in a way that might suggest shelters of the form observed at Morton A. In some of the areas of the midden discrete episodes of activity or deposition could be noted: for example a deposit of four bone tools, three poorly worked cores and a handful of flakes on a variety of material types, associated with the bones of red deer, cod, haddock, guillemot, cormorant and thrush. Several of these short-term events were recorded, but one area seemed to have undergone more extended occupation. Here around a hearth setting 285 bones of mammals, birds and fish was uncovered, as well as a wide variety of shellfish (limpet, periwinkle, pelican's foot, necklace shell, dog whelk, common mussel, prickly cockle, venus shell, Baltic tellin, trough shell, common otter shell), including a pelican’s foot shell with a double perforation. Four bone bevel-ended tools were found and a small assemblage of 72 stone tools, including manuports, a core, chopping tools and flakes. Thirty-five of these were unworked nodules of chalcedony, 17 of which were scattered over the area of a square metre and may represent a cache. Further evidence for intermittent use of the midden comes from seasonality studies (Deith 1983). These suggest the majority of the shells were collected during the winter, but that there was occasional summer activity and rarer late summer/autumn visits. Shellfish gathering may have been embedded within trips to procure lithic material along the beaches (Deith 1986). 8 Western Scotland Understanding of fifth millennium life in western Scotland has been dominated by the evidence from the Oronsay middens, particularly Cnoc Coig. Until recently there has been relatively little to compare with this, the large surveys of the Southern Hebrides Mesolithic project for example, failing to locate any substantial contemporary Mesolithic site (Mithen 2000). This has begun to change in recent years with excavations of small middens at cave sites in the region of Oban (see textbox 6.3) and on Ulva (Bonsall et al. 1994) and renewed excavations at Risga (Pollard et al 1996). Recent fieldwork has also located sites of this date on a number of other islands. In the north, on Lewis, dates of 4400-4000BC have been obtained from Tráigh na Beirigh 1. This site is a shell midden (figure 6.6), overlying a buried soil which also preserves evidence for late Mesolithic activity (Piper and Church 2012). Further south new sites have also been located on Colonsay and Islay. These are all islands where Late Mesolithic evidence has been recorded. In the far north, visits to Shetland, dating to the last century or so of the Mesolithic have been recorded from West Voe, where a midden, that continued to be visited in the Neolithic, first began to accumulate. The lowest layer of the midden is composed of oysters, with an overlying layer comprising of limpets, seal and sea-birds belonging to the period of the transition (Melton 2009). <insert figure 6.6 here> Figure 6.6. Eroding section at Tráigh na Beirigh 1 prior to excavation (copyright Peter Rowley-Conwy) <start textbox here> Textbox 6.3. The ‘Obanian’ sites In the last decades of the nineteenth century several shell middens were excavated on the small island of Oronsay and in caves and rockshelters on the mainland in the vicinity of Oban. These sites were associated with bone and antler artefacts, in particular biserial barbed points (figure 6.7) and bevel-ended tools (figure 4.8), some of the latter were also made from stone. Small flaked stone assemblages were present but microliths, and often retouched pieces more broadly, were entirely lacking. Given their close association with the beaches of the post-glacial maximum transgression, this combination of artefacts was seen as indicative of a geographically restricted, temporally-discrete cultural grouping. It was named the ‘Obanian’ by Movius (1940) who compared it with the non-microlithic late Mesolithic in Ireland, known as the Larnian. <insert figure 6.7 here> Figure 6.7. Barbed points from Druimvargie (1-2) and Caisteal nan Gillean I (3-5) (Anderson 1898) The concept of the Obanian has been dismantled in recent years, as suggestions that the differences in material culture between Obanian and microlithic assemblages may be functional rather than cultural have gathered momentum (Bonsall 1996). Radiocarbon dating programmes focused on organic artefacts have revealed that the midden sites do not occupy a discrete temporal span, but in fact include some of the earliest dates from the west coast (at Druimvargie) as well as some of the latest (Cnoc Coig and Carding Mill Bay). At the same time, dates became available for sites with microliths in the same region, indicating these were contemporary with the Obanian. Excavations at Lón Mór for example revealed microlithic sites and shell middens along the margins of a former marine embayment and 9 dates that span the late seventh to late fifth millennium, yet Lón Mór is less than a kilometre from Carding Mill Bay and Rascoille (Bonsall 1996, Bonsall et al. 2009). Furthermore, excavations at Risga, a geographic outlier but described by Lacaille (1954) as an Obanian site, have revealed a rich lithic assemblage in which microliths were well-represented (Polllard 2000), while blade debitage and a single microlith have been recorded at Ulva Cave (Bonsall et al. 1994). Site Date range Mesolithic age radiocarbon measurements Druimvargie 7570-6465 cal BC 7570-7175 cal BC (OxA-4608; 8340±80) 7040-6660 cal BC (OxA-4609; 7890±80) 7030-6465 cal BC (OxA-1948; 7810±90) Ulva Cave 6650-6265 cal BC 6650-6390 cal BC (GU-2600; 7655±80) 4770-4180 cal BC 6640-6265 cal BC (GU-2601; 7615±80) 4770-4455 cal BC (OxA-3738; 5750±70) 4705-4360 cal BC (GU-2602; 5685±80) 4545-4180 cal BC (GU-2603; 5525±80) Raschoille 6650-6020 cal BC 6650-6370 cal BC (OxA-8396; 7640±80) 5720-5010 cal BC 6590-6330 cal BC (OxA-8397; 7575±75) 4040-3780 cal BC 6440-6250 cal BC (OxA-3895; 7495±50) 6470-6215 cal BC (OxA-8398; 7480±75) 6350-5990 cal BC (OxA-8535; 7265±80) 6225-6020 cal BC (OxA-8439; 7250±55) 5720-5010 cal BC (OxA-8538; 6460±180) 4040-3780 cal BC (OxA-4838; 5115±55) MacArthur Cave 5730-5490 cal BC 5730-5490 cal BC (OxA-1949; 6700±80) Risga 5205-4555 cal BC 5205-4705 cal BC (OxA-2203; 6000±90) 4905-4555 cal BC (OxA-3737; 5875±65) Carding Mill Bay 1 4235-3655 cal BC 4235-3795 cal BC (OxA-3740; 5190±85) 3965-3715 cal BC (GU-2796; 5060±50) 3940-3655 cal BC (GU-2797; 4980±50) Casteal nan Gillean I 5320-4730 cal BC 5320-4940 cal BC (Q-3008; 6190±80) 4545-4000 cal BC 5290-4840 cal BC (Q-3007; 6120±80) 5205-4730 cal BC (Q-3009; 6035±70) 4545-4045 cal BC (Q-3010; 5485±110) 4500-4000 cal BC (Q-3011; 5450±110) Cnoc Sligeach 5220-3775 cal BC 5220-4055 cal BC (Gx-1904; 5755±250) 4770-3775 cal BC (BM-670; 5426±220) Priory Midden 4880-4595 cal BC (Q-3001; 5870±50) 4960-4450 cal BC (Q-3000; 5825±110) Caisteal nan Gillean 4725-3805 cal BC (OxA-8004; 5470±45) II 4455-4070 cal BC (Q-1355; 5460±65) 4725-3805 cal BC (Birm-347; 5450±195) Cnoc Coig (OxA-8004; 5470±45) 4680-4370 cal BC (Q-3006; 5675±60) 4650-4335 cal BC (Q-3005; 5650±60) 4765-4265 cal BC (Q-1353; 5645±110) 4830-3965 cal BC (Q-1354; 5535±195) 4550-4045 cal BC (Q-1351; 5495±110) (OxA-8014; 5495±55) 4690-3810 cal BC (Q-1352; 5430±180) Table 6.2. Radiocarbon dates for ‘Obanian’ sites 10 Amongst the ‘Obanian’ sites, the cave site middens in the Oban area and Ulva Cave to the north, have a wider date range than the Oronsay middens, mainly spanning the late eighth millennium to the end of the sixth millennium and are discussed in the previous chapter. Only Raschoille and Carding Mill Bay may have very late Mesolithic dates (table 6.2), one from the latter centring on 4000BC on an antler bevel-ended tool (Ashmoor 2004). Both sites show early Neolithic activity and some of these potentially very late Mesolithic dates could reflect the activities of the earliest Neolithic inhabitants of the region. <insert figure 6.8 here> Figure 6.8. Distribution of Obanian sites Beyond Oban, there is good evidence for fifth millennium activity on other sites previously seen as Obanian where recent excavations have also taken place. These excavations, at Ulva and Risga, have also been instrumental in showing the association of microliths with Obanian sites. At Ulva Cave, on Ulva Island, just off Mull, a midden 8m across and 0.35m thick was located in a large cave measuring 17 by 15m (Pickard and Bonsall 2009, Russel et al. 1995). The midden was composed predominantly of limpets and to a lesser extent periwinkles, with dogwelk also present. Limpets were gathered from the middle and lower reaches of a rocky shore. Crabs were exploited for food and may have been collected in the summer months (Pickard and Bonsall 2009). Mammal and fish bones, along with the occasional hazelnut shells were also incorporated (Bonsall et al. 1994, Russel et al. 1995). The midden seems to have had a long duration, with the lower layers accumulating in the mid-seventh millennium and the upper layers in the mid-fifth, with evidence too for early fourth millennium and early Neolithic activity. Organic artefacts include an antler bevel-ended tool and a perforated cowrie, though these appear much rarer than at other middens (Bonsall 1996). A single microlith was recovered amongst a small lithic assemblage. Northernmost of the ‘Obanian’ sites is a midden on the island of Risga, a small rocky outcrop in the middle of Loch Sunart, excavated in the 1920s with minimal records. The midden was on a rocky platform at 9m asl. At the time of occupation in the fifth millennium BC, this would be not far above the spring high tide mark (Bradley et al 2012). The midden was 30cm thick and in the centre a layer of burnt stones and several hearths was recorded (T. Pollard 1996). In contrast to the middens in the Oban area, a large lithic assemblage was recovered during the 1920s excavations. The midden seems to overly a ‘sooty layer’ and there have been suggestions that there was an earlier occupation focused on microlith production, followed by an ‘Obanian’ midden lacking these tools (Woodman 1989). However, examination of records of the 1920s excavations suggests greater stratigraphic complexity and that the midden and sooty layer probably accumulated through a number of different events. At times the debris of activities taking place in the ‘sooty’ layer may have been discarded in the midden (T. Pollard et al. 1996). The midden was dominated by limpet with fish and birds well-represented and a variety of terrestrial and marine mammals (Lacaille 1954) (table 6.4-6.7), exploited for food, fur and raw materials. A wide variety of organic tools were recovered, including bevel-ended tools, barbed points, simple bone points and a T-shaped antler axe. Stone bevel-ended tools were recovered, as was a large assemblage of over 18,000 lithic artefacts, made mostly on quartz, but with quantities of flint and bloodstone (Pollard 1990). Perforated cowries and pigment add to the range of material recovered (Lacaille 1954, 233). A range of relatively amorphous tools were present, including heavy duty picks. Further excavations by Banks and Pollard in the 1990s discovered further occupation beyond the midden, including several hearths, pits and curvilinear slots, accompanied by a lithic assemblage with numerous microliths (Atkinson et al., 1993, T. Pollard et al. 1994, Banks and Pollard 1998). 11 Dates from a bevel-ended tool and the T-shaped axe indicate occupation in the first half of the fifth millennium BC. These late dates are probably representative of the Mesolithic use of the midden; any earlier occupation is likely to have been destroyed during the maximum post-glacial transgression. Neolithic pottery is also present (T. Pollard 1990) and without further dates the extent of activity of various periods cannot be determined. <end textbox> Midden Location Type Length Breadth Height (m) (m) (m) Sand Applecross Rockshelter 8 8 1 MacArthur Cave Oban Cave <7.5 <6 0.9 Druimvargie Oban Cave 4.5 3 1.2 Distillery Cave Oban Cave <3.5 <2.7 ? Risga Loch Sunart Open air 0.3 Caisteal nan Gillean I Oronsay Open air 45 45 2.4 Caisteal nan Gillean II Oronsay Open air 15 15 0.6 Cnoc Sligeach Oronsay Open air 90 50 1.7 Priory Midden Oronsay Open air 25 25 0.8 Cnoc Coig Oronsay Open air 25 20 0.7 Port Lobh Colonsay Open air 30 25 Table 6.3. Features of west coast middens On Oronsay The large shell middens found on the small island of Oronsay (figure 6.9) have been seen to typify fifth millennium settlement on the west coast, though in many ways, at least until the location of a similar sized midden at Port Lobh (Finlay et al 2019), they seemed rather unique. It has been suggested that Oronsay was occupied by sedentary, marine-adapted hunter-gatherers (Mellars 1987); however even the findings of the original excavation indicated that the people visiting Oronsay were sea-farers who travelled widely and more recent work has further emphasised the complex settlement pattern in the region. While today the island can be reached at low tide from the larger island of Colonsay, in the late Mesolithic, with higher relative sea-level, boats would have been needed. Estimates of the size of the island at the time of occupation suggest it would have been only in the region of c4km2 (Mellars 1987). <insert figure 6.9 here> Figure 6.9. Aerial view of Oronsay. Copyright NCAP/ncap.org.uk The Oronsay middens are considerably larger than the small cave-middens of the mainland (see table 6.3) and appear even more prominent because of deep deposits of wind-blown sand (Warren 2007). In the Mesolithic these accumulations of gleaming white shells would have been visible markers when approaching the island by boat (Finlay 2004). These large mounds attracted the attentions of antiquarians who excavated at Caisteal nan Gillean I between 1879 and 1882, at Cnoc Sligeach in 1884 and 1913 and at a third site Cnoc Riach, whose location is now unknown, but may possibly be Cnoc Coig. Cnoc Coig also underwent excavation in 1913 (when it was known as Druim Hastell or the Viking Mound). Further work 12 was carried out in the 1970s when Paul Mellars undertook extensive excavations at Cnoc Coig and more limited testing of the other middens. Most middens are on the eastern side of the island, facing towards Jura. Cnoc Sligeach is northernmost and Caisteal nan Gillean I southernmost. Only the Priory Midden is on the western part of the island, facing the empty expanse of the Atlantic Ocean. The shell middens represented a fixed point in a shifting, tidal landscape, albeit one threatened by spring tides and storm surges. Tony Pollard (1996) has drawn out the temporality of activities on these shorelines. The rhythms of low tide brought people down to the sea strand to harvest shell-fish and check fish traps at times dictated by the pull of the tides rather than the movement of the sun. On a broader scale, travel between the islands was affected by the seasons and patterns of weather. Reading the seascape was an important skill, particularly as it seems the last centuries of the Mesolithic represented a period of increased storminess (Garrow and Sturt 2011). The excavated middens share a number of common features. They are composed of shells, amongst which limpets dominate; edible periwinkle are common and dog welk well- represented (table 6.4). Fish bones, often saithe, are common (table 6.5). Seasonality evidence, derived from the study of saithe otoliths, suggests fishing activities took place at different times of the year at the different middens (Mellars and Wilkinson 1980). At Cnoc Coig, where sampling was most extensive, fishing, both during a pre-midden occupation and during the accumulation of the midden itself, seems to have taken place mainly in the autumn, though occasional mid-summer and winter events were also recorded. At Cnoc Sligeach the evidence suggests summer, and at Priory Midden and Caisteal nan Gillean II, winter, though it should be noted that these results come from a single column sample taken from very large sites. Caisteal Caisteal nan nan Gillean Cnoc Priory Cnoc Port Risga Gillean I II Sligeach Midden Coig Lobh Limpet +++ +++ +++ +++ +++ +++ +++ Edible Periwinkle + + + + + + +++ Flat periwinkle + + Cowrie + + + Dog whelk ++ + ++ + + Whelk + + + + Mussel + + + + + Oyster + + + Scallop + + + + + Cockle + + + + Topshell + + Razor clam + + + Crab + + + + + Sea urchin + Table 6.4. Shellfish from fifth millennium middens Caisteal nan Cnoc Cnoc Port Risga Gillean I Sligeach Coig Lobh 13 Tope + + Dog fish + + + Ray + Angel-fish + Skate + + Conger eel + + Saithe +++ +++ Cod + Haddock + Sea-bream + Black bream + + Wrasse + + Mullet + + Norway pout + Table 6.5. Fish remains from fifth millennium middens. Full details are not available for all sites. Sixty percent of the mammalian remains at Cnoc Coig come from seals, almost all grey seals (table 6.6). Grey seals – both adults and pups – were most likely killed in the autumn when they hauled themselves onto rocks to calve. Entire seal carcasses seem to have been brought to the middens by boat. The remains of whales and dolphins or porpoises were also found. A worked whale bone was recovered next to a hearth at the Priory Midden (Mellars 1987) and a whale bone point was recovered from Caisteal nan Gillean I (Lacaille 1954). Otter bones were common, outnumbered only by seals. Otters were brought to the site complete for skinning; they do not seem to have been eaten, instead, their bodies seem to have been thrown on the fire (Grigson and Mellars 1987, 278), perhaps a culturally appropriate way of dealing with a significant animal. Terrestrial mammals are restricted to red deer and pig, both seemingly more important as raw materials than food. Red deer remains consist mainly of antler, imported to make tools such as mattocks, bevel-ended tools and awls. Lower limb elements of both red deer and pig are well-represented: these too seem related to the production of bevel-ended tools (Grigson and Mellars 1987). Caisteal Caisteal nan nan Gillean Cnoc Priory Cnon Port Species Risga* Gillean I II Sligeach Midden Coig Lobh Red deer + + ++ + + ++ + Pig + + + ++ Badger + Otter + + + + ++ + Pine Marten + + Common seal + + + + Grey seal + + + ++ +++ Whale + + + Dolphin + + Table 6.6. Mammalian remains from fifth century Scottish middens 14 An extremely wide range of birds have been recovered from the middens, particularly Cnoc Coig (table 6.7). These are mostly pelagic sea birds, particularly the great auk, which makes up 15% of the birds at Cnoc Coig. As auks, puffin, guillemot and razorbill only come ashore to breed, this indicates focused fowling in the summer months, though there are also a small number of winter visitors such as the little auk (Best and Mulville 2016). Waterbirds such as duck, geese and swans are well-represented and a small number of moorland and woodland birds are present. Plant foods are represented by large quantities of charred hazelnut shells (Mellars and Wilkinson 1980). Caisteal nan Cnoc Cnoc Risga Gillean I Sligeach Coig Swan + ++ Goose + + ++ Duck ++ Red-breasted Merganser + + Cormorant/shag + + ++ Gannet + + ++ Plover + Tern + + + Gull + + + Razorbill + + + ++ Guillimot + + + ++ Great Auk + + + +++ Little auk + Puffin ++ Water Rail + + + Curlew + Sparrowhawk + Buzzard + Quail + Snipe + Woodcock + Blackbird + Raven + Table 6.7. Bird remains from fifth millennium middens The middens were places focused on cooking and processing: Burnt stones, small pits and hearths are extremely common. Large quantities of tools have been found on the more extensively excavated middens (see table 6.8): At Cnoc Coig, 436 bone and antler and 359 stone bevel-ended tools were found (Nolan 1986); on Caisteal nan Gillean I, 150 organic bevel-ended tools, 10 biserial barbed points (see figure 6.7, 3-5), numerous simple points (including one made from whale bone) and eight T-shaped antler axes. In contrasted chipped stone is less common than typical on other Mesolithic sites: assemblages consist mainly of small flint flakes, detached mainly using the bipolar (anvil) technique. Some middens are associated with larger assemblages than others but retouch is rare in each case: retouched pieces are absent at Cnoc Sligeach, with only a handful recorded at Caisteal nan Gillean I (Lacaille 1954). At Cnoc Coig, bladelet debitage occurred, particularly in the later phases of the midden and some formal tools are present (Pirie et al 2015). Shells 15 were not just food waste but material culture, with perforated cowries worn as beads, scallops used as scoops and larger shells used as containers: a cache of 16 unperforated cowries were found in the shell of a prickly cockle (Mellars 1987). Site Bone/antler Stone bevel Barbed Antler Lithics bevel ended tool ended tool point mattock MacArthur Cave 140 0 7 0 20 Druimvargie 18 4 2 1 Risga P P P 4 18,000 Caisteal nan Gillean I 150 210 10 9 90 Cnoc Sligeach 36 150 7 3 Rare Cnoc Coig 436 359 2 11 Table 6.8. Material culture from ‘Obanian’ midden sites A feature of three of the middens is the presence of human bone. Fifty-five human bones were recovered, the majority (49) from Cnoc Coig, five came from Caisteal nan Gillean II and a single phalanx from the Priory Midden. The majority of these are hand and foot bones, though 16 represent other elements, mostly bones of the torso. Of limb bones only a single tibia fragment was found. The bones range in age from those of a child to an adolescent, to those fully adult, and both men and women seem to be present, suggesting age and gender were not a factor in the nature of funerary treatment. A minimum of seven individuals are represented (Meiklejohn and Denston 1987). The unusual patterning of elements has led to suggestions that human remains were excarnated on the middens (ibid.). Bodies may have been laid out on the middens themselves or scaffolding above it, with the larger parts disposed of elsewhere – perhaps into the sea (T. Pollard 1996) – and the smaller parts overlooked. However, analysis of the spatial location of the human remains suggests that rather than being forgotten, they played a more active role on the site (Meiklejohn et al 2005). A cluster of hand and foot bones seem to have been placed on a seal flipper (Nolan 1986, Finlayson 1996). Several human elements seem to be associated with the two structures on the site, suggesting some were curated. Nolan (1986) also notes the similarity in skeletal part representation between humans and red deer and pig, all three of which are represented by lower limb elements. This pattern in the animals has been explained through the use of lower limb bones for tools, and it should be noted that while finger and toe bones are common, the larger human limb bones, of suitable size for tool manufacture, are absent. Cnog Coig is the site where the most extensive fieldwork has taken place; here196 square metres, around 70-75% of the entire midden, were excavated during the 1970s excavations (figure 6.10). The site has been the subject of extensive spatial analysis (Noland 1986) and is also the best dated (Wicks et al. 2014). As such the archaeology of the midden is worth exploring in some detail to understand the process of midden accumulation and how this product of ephemeral activity and discard practices emerged as a social space; perhaps even, as has been suggested, as a symbolically-charged Mesolithic monument (T. Pollard 1996, J. Pollard 2000, Cobb 2005). <insert figure 6.10 here> Figure 6.10. Plan of midden and features at Cnoc Coig (redrawn after Mellars 1987) 16 At the time of occupation, Cnoc Coig would have been close to a receding sea, lying on an area of dunes. Sedimentological analysis suggests the midden lay close to the junction of the beach and the dune system. A bank of sand to the north would have provided shelter, and a similar bank may once have existed to the south also. Before the midden started to accumulate, there is evidence that people had already visited the location. A layer of debris was recorded on a weak soil horizon, indicating the dune system underwent a period of stabilisation during which vegetation grew. This early occupation in the northern area of the site, in the lea of the land bank, was small, spread over a 5m by 4m area. It consisted of two adjacent hearths, only three metres apart, marked by the presence of burnt stone, burnt shells and charcoal. Small piles of shells surrounded the hearths. Beyond these, shells were sporadic but there were patches of fish bone, all young saithe. Artefacts consisted of an anvil stone, three bevel-ended tools and the base of a shed red deer antler (Mellars 1987). This occupation, which Mellars suggests was a single event (ibid., 232) probably occurred sometime in or after the 46th or 45th century BC (Wicks et al. 2014). Above this episode of pre-midden activity 20-30cm of sterile wind-blown sand accumulated, indicating the resumption of dune formation. Within this was a thin palaeosol, indicating some gap between this early occupation and the start of midden accumulation, though wind- blown sand can accumulate rapidly and a palaeosol may develop with a few years of stabilisation. The earliest dates on the midden are very similar to those from pre-midden activity, albeit all have large standard deviations and are on bulked, unidentified wood, so should be considered termini post quos for these events. The extensive excavations at Cnoc Coig and the detailed spatial analysis of the midden undertaken by Nolan (1986), give an insight into the accumulation of the midden and how its spaces were produced. As McFadyen (2006) argues, in the Mesolithic it is activities and the deposition of the generata of ephemeral activities that make space, and nowhere can this be seen more clearly than at Cnoc Coig. Activity seems to have started with a small semi- circular structure, marked out by stakeholes (structure 1), with a diameter of 3-3.5m (Mellars 1987, 238). A midden (the phase 1 midden) began to accumulate as food waste and hearth debris from the structure was piled up behind it and propped against it. This midden was also a focus for activities with three hearths found at different levels within it. Midden material also began to accumulate slightly further to the south (phase 2 midden). Refits of human bone between the structure and the lower part of this midden suggest it was also accumulating while the structure was in use, and this lower part (2a) may be contemporary with the phase 1 midden (Mellars 1987, 226). In the centre of structure 1, a pit was dug, which contained a bevel-ended tool and a red deer bone which may represent a foundation deposit (Nolan 1986, McInnes 2015). A hearth was laid on top of this pit. McInnes (2015) has drawn attention to the importance of hearths and the transformations enacted by fire at Cnoc Coig, She notes particularly the importance of this hearth as a central focus, which was repeatedly cleared out and remade in the same spot: Mellars (1987, 238) notes the presence several superimposed layers of burning, as well as dumps of burnt material in the surrounding midden. In addition to the feature underlying it, the hearth was surrounded by a number of enigmatic pits containing midden material: As Mellars states: ‘The interpretation of these in functional terms remains problematic’ (ibid, 238). Two of these pits contained burnt stones – these ‘stone holes’ were a common feature of the midden architecture. These were not post holes with post-packing, but instead seem to be related to cooking and heating (ibid 240). They echo the late Mesolithic pattern of incorporating burnt and transformed materials, often related to particular acts of food preparation, into pits. Burnt material, cleared from hearths, was also found as layers in the midden. 17 The structure is associated with the deposition of both animal and human remains. The only two barbed antler points from the site were both found on its edge. Near the base of the structure were four human phalanges (Nolan 1986). One of the phalanges refits with another found in the phase 2 midden (Meiklejohn et al. 2005). In total a grouping of 16 human bones, mainly phalanges and metacarpals/tarsals, have been found in this area of the site, with at least three individuals represented. This group is found in three main areas: In the lowest 10cm of the phase 1 structure, relatively high (30-40cm) in the phase 1 midden and at mid- level (25-30cm) in the phase 2 midden, below a sand lens that is an important stratigraphic marker in this area (Nolan 1986). While it is possible that the human bone within the structure has slumped in when walling collapsed or was removed and thus derives from excarnation on the midden, it could also suggest curation and circulation of some of the human remains from Cnoc Coig. The pattern of human bone deposition suggests the earliest parts of the phase 2 midden, located just to the east of the phase 1 structure, are contemporary with the phase 1 midden. However deposition into the phase 2 midden continued above a sand lens that marks the end of the phase 1 midden; this lens might suggest a short period of abandonment. This phase B midden started with the deposition of cetacean bone. It also contained clusters of seal bones and a scatter of human bones found close to the remains of red deer, otter and a second cetacean (Nolan 1986, figure 67). Bird remains are common in the phase 2 midden and include discrete clusters of quail, goose, teal, duck and swan bones which are rare elsewhere on the site. Bevel-ended tools were also found throughout this midden. The western area of the site was not used for middening during the early phases of occupation. This instead seems to have been place where activities took place around open- air hearths. A major hearth, one of several at the lowest levels in this area, was associated with clusters of burnt stones and a number of stone holes, perhaps indicating large scale cooking activities (Mellars 1987, figure 14.18). Antler is also common in these lower levels (Nolan 1986), perhaps indicating that this area was also a focus of craft activities. Phase 3 marks a shift in the spatial focus of the use of the site, which moved several metres to the south. This involved the building of a new structure, very similar in form to that of phase 1: A semi-circular arrangement of stakeholes, approximately 3.5m in diameter and with a central hearth. This hearth similarly showed evidence for many superimposed firings with a thickness of 25cm (Mellars 1987). Antler working seems to have taken place around the hearth (Nolan 1986). Ochre is also common within the structure. Middening recommenced, with the main focus immediately to the northeast (in an area previously associated with cooking fish) and also to the east and south of the structure; the midden, as in phase 1 encircling the structure and perhaps eventually merging with/forming its walls. This phase 3 midden may also have included material redistributed from the earlier phases (Meiklejohn et al. 2005); it may have been important that the new structure was surrounded by evidence of past events, connecting people to the activities of past generations. A second major cluster of 18 human bones comes from the area of the phase 3 structure, most found between c.10 and 30cm from the base of deposits, which is at broadly the same level as the main hearth associated with this building (Nolan 1986, Meiklejohn et al. 2005). This cluster may even lie below the hearth (Nolan 1986, 254) and thus might represent a foundation deposit. The human remains are strongly associated with a cluster of ochre. There are refits between this group and two outliers, found in midden material to the south of the structure. Two radiocarbon dates are available for human remains in this cluster: 4250- 3840 cal BC (OxA-8014; 5465±55, marine correction applied) and 4330-4000 cal BC (OxA- 8019, 5615±45, marine correction applied). This group also consists predominantly of bones 18 from the hands and the feet and represents at least three individuals. A collection of these bones was found immediately above a cluster of bones from a seal-flipper. The fact that both major groupings of human bones are strongly associated with the only two structures on the site may be significant. It is difficult to be completely certain whether these are contemporary with the use of the structure or date to the period following their abandonment, though from their stratigraphic position the former seems much more likely. If this is the case, there remains the possibility that the buildings themselves were mortuary structures, or that the parts of the dead, excarnated on the middens were brought within the domestic sphere to co-exist with the living. The phase 3 midden deposits contain three well-defined extensive dark layers, comprising burnt shell, fish bones and charcoal, which seem to represent debris cleared out of hearths, probably from the adjacent structure. Deposition commenced with two cetacean bones and the upper levels contain the remains of otter, pig and red deer. Oysters are well-represented in comparison with other areas of the site. Antler waste material and foetal and young seal bones are particularly common in the midden to the southeast, and three antler mattock fragments were found in the midden to the east of the structure. The western area seems to have seen lower levels of use throughout the time Cnoc Coig was visited. After being used as an area for activities focused around hearths, shell began to be accumulated, but at low levels, interspersed by sand. The emplacement of hearths and antler working continued in this area during this phase and bevel-ended tools are also common. Mellars has suggested that much of the shell accumulation in this area seems to be fairly late in the sequence, equivalent to phase 3, or even later. This has been confirmed by a recent radiocarbon measurement on pig, the only one from this area, which yielded a date of 3980-3805 cal BC (OxA-29937, 5122±30BP) (Charlton et al. 2016). This time period would traditionally be associated with the presence of Neolithic groups, at least in other parts of Britain. During the later occupation of the site, the area seems to have been used for specific tasks, particularly involving processing of animals: much of the seal bone from the site clusters in this area and there are several small scatters of otter, pig and red deer bone in the upper levels. While less is known about the generation of the other middens on the island due to the smaller scale of excavations at these sites, there are hints of similar processes. At Caisteal nan Gillean II a hollow 30-40cm in depth was noted below the midden layers, associated with a posthole to the east, which may have represented a structure (Mellars 1987). This may suggest that here too, the midden built up around a structure. A large pit was also recorded at the Priory Midden. At Cnoc Sligeach, the largest of the Oronsay middens, material seems to have accumulated in two or three distinct centres which, over time, merged to create a single mound, as at Cnoc Coig (Mellars 1986, 195). Two distinct layers were noted. The earliest was characterised by low density shell debris associated with wind- blown sand, giving the impression of slow, sporadic accumulation of shells. This contrasts with a higher density of shells and occupation debris, including much burnt material, probably cleared from hearths elsewhere on the site, indicating different rhythms and intensities of occupation. Tony Pollard (1996, 203) argues that these middens may have been liminal places, located just beyond the margins of the tide, where the resources of both the land and the sea were processed. He points to the presence of marine mammals on the middens, animals that belong to both land and sea. The presence of the dead on the middens may also support ideas these were liminal spaces. Excarnation is a process of liminality, transforming known individuals into clean bones. Cobb (2005) however argues that liminality may be a less 19 useful concept for the strandloopers and marine voyagers who habitually traversed these zones, preferring to see the middens as places of transformation, where people were remade into ancestors and fish, and shell-fish into food (T. Pollard 1996, Cobb and Gray Jones 2018). Key transformations were enacted at Cnoc Coig, which may have made the middens powerful accumulations. They reflect a long-standing interest shown by Mesolithic groups in differential treatment of hearth debris and other things transformed by fire. They were places where fish, shell-fish and marine mammals were transformed into food and the dead transitioned from people to bone. The dead may have undergone further transformations: The similarity between certain seal and human remains in death was recognised by those that visited Cnoc Coig and placed human phalanges on top of a seal flipper (Nolan 1986, Conneller 2006). If the majority of the human bones from people excarnated on the midden were deposited in the sea, as suggested by Pollard (1996), this similarity may be more pronounced. Perhaps the seals were understood carry the souls of the dead, like the selchies seen as lost human souls in Scottish oral tradition; Radovanovic (1997) has made a similar argument for the role of migratory sturgeon at Lepenski Vir. She suggests they were seen to carry the souls of the dead downstream in the autumn, returning in the spring. The occupation at Cnoc Coig coincides with the season when grey seals come to shore to give birth, perhaps conceived as the return of human souls, in a continuing cycle of life. Tony Pollard (1996) makes the point that as places of excarnation the middens may have played the same role as Neolithic chambered tombs (see also Thomas and Tilley 1993). This would also extend to a role as a landscape marker; they would be visible as gleaming white mounds to those approaching the island by boat, memorialising past activities (Finlay 2004), including perhaps significant events such as feasts (Thomas and Tilley 1993). Their location may have been a way of broadcasting ancestral connections between Mesolithic groups and a place that was also important for access to the sea (T. Pollard 1996). What relationship the people who made the middens had to sites beyond Oronsay has been the matter of some debate. Mellars, perhaps influenced by a prevailing interest in complex hunter-gatherers, for whom a key indicator was sedentism, linked the varied seasonality indicators from the different middens to suggest people were present on the island throughout the year (Mellars and Wilkinson 1980). Isotopic analysis of the human remains was seen to support this, as five samples of human bone from Cnoc Coig indicated a diet focused almost entirely on marine resources (though a single sample from Caisteal nan Gillean II showed a mixed diet) (Richards and Mellars 1998). However Grigson’s faunal analysis showed people had access to red deer, which were unlikely to be present on such a small island as Oronsay (Grigson and Mellars 1987). Two sizes of red deer were represented: larger animals, probably obtained from the mainland, and smaller sized animals which were likely to be have been adapted to the more restricted resources of one of the larger nearby islands. Such evidence, as well as increasing knowledge of the Mesolithic of nearby islands, has led others to argue that visits to Oronsay were more sporadic, and that the island was just one of a network of places visited by hunter-gatherers who focused on marine resources but also exploited the flora and fauna of adjacent areas (Mithen and Finlayson 1991, Mithen 2000b, Wicks et al. 2014). While the surveys of the Southern Hebrides Mesolithic project failed to locate any sites contemporary with Cnoc Coig, more recent work has uncovered fifth millennium sites on both Colonsay and Islay, showing a variety of settlement forms in the region (Wicks et al. 2014, Finlay et al. 2019). Textbox 6.4: T-axes 20 Over the past few decades, several dating programmes, combined with typological study, have elucidated changing technological traditions of antler tool production (Smith 1989, Elliott 2015). Elliott has noted that two main types of antler digging or chopping tools, termed mattocks by Smith (1989), were in use during the fifth millennium. These axes are tools made on the beam of the antlers of red deer (elk, whose antlers were used for early Mesolithic mattocks were seemingly extinct by this time). The initial associations of these tools, for example at Meiklewood (see above), suggested they may have been used as blubber hammers for exploiting whales (Clark 1947). Smith (1989), following Clark’s interpretations of the much earlier elk antler examples from Star Carr, suggested they could have been used for digging, leading to their association with plant exploitation (Zvelebil 1994). Elliott (2015) suggests they may have played a more diverse role, with woodworking and animal butchery also possibilities. The first of Elliott’s two late Mesolithic mattock groups can be defined as having a perforation through the beam itself; they have a longer currency, appearing first around 7000BC, with dated Mesolithic examples extending to the end of the period. Amongst the second group the perforation is made through the stump of the removed trez tine, giving them, when perforated, a T-shaped profile (Elliott 2015). These T-axes are much more temporally restricted: two, from Meiklewood (figure 6.11) and Risga, have direct dates and belong to the first few centuries of the fifth millennium (Tolan-Smith and Bonsall 1999). A third comes from the Priory Midden on Oronsay and is undated, but the midden itself has dates falling into the second half of the fifth millennium (Wicks et al 2014). Two new finds from Tarradale (figure 6.5) come from a midden with fifth millennium dates. <insert figure 6.11 here> Figure 6.11. T-axe from Meiklewood. Copyright Ben Elliott These Scottish T-axes are part of a class of objects that have a widespread distribution across Europe, from the Iron Gates in the east, through central Europe to the Baltic coast, where they are classically associated with the southern Scandinavian Ertebølle. In continental Europe they extend as far west as Belgium (Elliott 2015, figure 11). In these contexts too they tend to have fifth millennium dates and as such are associated variously with Mesolithic groups and pottery using hunter-gatherers. Mesolithic Britain has for a long time been seen as isolated following the inundation of Doggerland (Jacobi 1976). The connection of the Scottish finds with this broader group suggests this is not the case, and that parts of Britain at least were caught up in wider networks of marine voyaging. Colonsay and Islay Until recently there was little trace of occupation from the last 1500 years of the Mesolithic on either Islay or Colonsay, the location of extensive seventh and sixth millennium Mesolithic occupation (see chapter 4). This changed with the excavations of a late fifth millennium shell midden at Port Lobh on Colonsay (Finlay et al 2019) and an occupation site at Storakaig on Islay which has a series of radiocarbon dates spanning the second half of the fifth millennium BC and the first half of the fourth (Wicks et al 2014). The midden at Port Lobh (figure 6.12), measuring 30 by 25m, demonstrates that large shell middens in western Scotland are not confined to Oronsay. The shells would have stood at around the level of the main postglacial transgression, a few metres above the reach of high tides, on what would have been an extensive lagoon and estuary. This is a sheltered area, close to a freshwater stream. Charcoal is compatible with a local vegetation of birch-hazel scrub with willow and alder fringing the margins of the lagoon and patches of oak woodland on drier ground. 21 <insert figure 6.12 here> Figure 6.12. Port Lobh from the southeast (copyright Nyree FInlay) Six small testpits were excavated into the midden (Finlay et al. 2019). Though truncated by later activity the midden still stood 0.45m deep in places and was composed of shell, charred hazelnut shells, burnt and unburnt bone, flint and fire-cracked rocks (figure 6.13). Limpets are the most common shell recovered, followed by periwinkle. These can be gathered from rocky shorelines close to the site. The large numbers of immature limpets suggest they may have been gathered as fishing bait, rather than for direct consumption; mature periwinkles are more common suggesting they were eaten. Though rocky shore species dominate, the more fragile-shelled sandy shore species are present in small numbers and include oyster, razor shell and mussel. Saithe are the dominant fish with smaller numbers of cod and a single otolith of Norway pout which was probably an incidental catch. Measurements of saith otoliths suggest autumn or winter fishing. These young saith were probably caught from rod and line fishing from the shore or with line or nets in shallow water. Larger cod suggest some offshore fishing. <insert figure 6.13 here> Figure 6.13. Midden deposits at Port Lobh (copyright Nyree FInlay) Radiocarbon dates indicate at least two phases of activity, one centred on the 43rd century cal BC and the other on the 42nd century, though the midden could also be the product of a larger number of small-scale intermittent visits. In places two phases can be discerned, suggesting changing activities: a lower layer was associated with the processing and cooking of large cod. Hazelnut shells were absent and the only charcoal was Maloideae. The upper layer, the main midden, has only small cod and debris from a more varied suite of activities. More ephemeral occupation evidence is also present on Colonsay, located away from the coast. At Staosnaig, a single large pit with stones, lithics and charred hazelnut shells has a very late Mesolithic date of 4360-4055 cal BC (5415±60BP; AA-21621). This is associated with a microlith assemblage dominated by straight backed bladelets, raising the intriguing possibility that final Mesolithic ‘rod’ assemblages (see chapter 1) may also be present in Scotland, though there are suggestions that the lithic material in this feature may be residual (Finlay et al. 2019). A more extensive inland occupation has been located on Islay, underlining that Final Mesolithic sites in the region were not exclusively located on the coast. A site at Storkaig, at 110m asl, in the centre of the island, consists of a dark organic horizon containing lithics, charred hazelnut shells and burnt bone around 18x13m in extent and sealed by peat (Wicks et al. 2014). Though burnt, some of the bone has been identified as red deer, roe, pig, badger and a small dog or fox. Bird, amphibian and fish remains are also represented. This appears to represent an organic midden deposit that spans the Mesolithic-Neolithic transition, and this continuity in occupation echoes the evidence from the small shell middens of Raschoille and Carding Mill Bay. Southwest Scotland/Northwest England In southwest Scotland and Cumbria there are a handful of sites dated to the fifth millennium. Those that do exist come from locations previously occupied during the late Mesolithic perhaps suggesting some continuity of landscape practices. Only 600m from the late 22 seventh millennium site of Littlehill Bridge in Ayrshire (see chapter 5), is Gallow Hill, where a charred hazelnut shell from a pit containing lithics and charcoal has been dated to 4800- 4550 cal BC (5835±35BP; GU-9806). This pit is one of two associated with lithic scatters indicating repeated occupation during the Mesolithic, focused on a small pool fringed with alder. The lithic assemblage includes a small amount of Arran pitchstone; Gallow Hill is close to the coast, and a 26km journey to Arran by boat (Donnelly and Macgregor 2005). In Dumfries, Barsalloch, like Low Clone (see chapter 5), is on Luce Bay and sited on a late glacial raised beach (Cormack 1970). In a layer of sandy material below the topsoil were a series of stone settings and five pits forming an arc, some of which were filled with flint or stones and which have been interpreted as hearths. A small assemblage of 461 flints was recovered from the small excavation, which includes eight microliths and several retouched pieces. A single radiocarbon date from unidentified charcoal probably indicates some of this occupation took place in the early fifth millennium BC. As at Low Clone all flint was obtained from local beaches and Cormack (1970, 78) noted the dominance of flint in this region of the coast, in comparison to sites to the east around the river Nith where flint, chert and quartz were all used. In the uplands of South Ayrshire around Loch Doon, Smittons T3 (see chapter 5) dates to the late fifth millennia BC (though this date may not be reliable), close to Smittons T1, a late sixth millennium site (Finlayson 1990). Further south at Eskmeals, a place used in the sixth millennium, fifth millennium BC activity is present at Williamson’s Moss. By this time the landscape had changed: A marine transgression created a gravel ridge, which allowed a pond to form. This became the focus for Final Mesolithic occupation activity, particularly in the later part of the fifth millennium (Bonsall 1989). The best evidence for the region comes from Stainton West (figure 6.14) on the river Eden in Cumbria (see chapter 5), where there seems to have been a gradual increase in the frequency of occupation, culminating the development of an intensely occupied and highly structured site by the middle centuries of the fifth millennium (figure 6.15). At the end of the sixth millennium, organic deposits began to form in the main palaeochannel, as the flow slowed and the water grew more stagnant. Here the remnants of the mid-sixth millennium beaver dam still seem to have had an effect on the hydrology as organic sediments built up around it, leading to ponding behind the structure. This seems to have attracted a family of beavers who constructed a lodge made from wood of oak, hazel, elm, alder and blackthorn in the first two centuries of the fifth millennium. <insert figure 6.14 here> Figure 6.14. Excavations at Stainton West, showing the palaeochannels. Copyright Oxford Archaeology Ltd. The late sixth/early fifth millennium organic deposits around the old beaver dam contained both lithics and worked wood, indicating human activity in the vicinity of the palaeochannel. A cluster of cores, flakes and pebble tools was found around the beaver lodge. Similarly, a small quantity of worked wood is also broadly contemporary with the beaver lodge, including a possible wooden artefact. Brown and colleagues (in press) suggests the clustering of material in the vicinity of the lodge may indicate it was re-used by people as a focus for activity, perhaps a launching spot for boats, or as bridge to cross the channel (Coles 2006). The evidence for activity in the channel is matched by evidence from the dryland area dating to the very end of the sixth millennium and the first three centuries of the fifth. A date of 5210-4930 cal BC (SUERC-42591) comes from charcoal associated with lithics in a later treethrow. Two further treethrows indicate occupation of the dryland in the first three centuries of the fifth millennium. One of these contains a significant amount of lithic material and thus may have been used for the deposition of midden material, though natural infilling 23 is also possible. A number of small lithic scatters, often associated with burnt flint that may indicate hearths, and often focused on microlith production and retooling may also belong to this phase. While this is still seemingly, as it had been in the sixth millennium, a place intermittently inhabited by humans and beavers, human visits now appear more frequent. <insert figure 6.15 here> Figure 6.15. Plan of excavations at Stainton West. Copyright Oxford Archaeology Ltd. Greater intensity of occupation is evident in the mid fifth millennium and the vast majority of the occupation, which generated 300,000 lithic artefacts, seems to belong to this phase, dated to between 4690 and 4490 cal BC. At this time, use of space at the site seems to have been highly structured, a characteristic of longer-inhabited settlements (Binford 1978b), with hearth features, dense areas of lithics and coarse stone tools and use of ochre. Lithic analysis has indicated a variety of more specialised areas (see figure 6.15). In the inhabitation area, three stake-hole structures were erected on slightly higher ground of the floodplain lying between the main palaeochannel and a cut-off meander which would have been a pool or boggy area. The first of these structures (structure 3) is defined by the pattern of debitage surrounding a space of c4m in diameter and seemingly respecting the adjacent midden. It is associated with a similar arrangement of hearth and earth oven as an earlier structure at the site dating to the sixth millennium (chapter 5). Structures 4 and 5, two features with partly overlapping footprints, are marked by arrangements of stakeholes. These were sub-circular features also around four metres across. These may also be associated with a similar hearth/cooking pit arrangement. To the north of the inhabitation area lay the site midden, an area seemingly with a long history that may have begun during the sixth millennium occupation of the site (chapter 5). Here were dumps of lithic material and burnt stone, the remnants of activities cleared from adjacent areas. There are some spatial patterns in the accumulation of midden material. The southern part of the midden has more microwear evidence for dry hide working, processing of wood and plants and discarded armatures. This is likely to reflect material dumped from the adjacent area of structures. By contrast the northern area of the midden has evidence for butchery and bone and antler working. This is likely to reflect dumping of tools used in the adjacent butchery location. The tool production area is characterised by more complete knapping sequences than found in the midden. Microliths were common, and a wide variety of other tools, such as awls, scrapers and retouched pieces were represented; it seems that tools were brought to this area for repair. To the northwest of the midden was another specialised area associated with a small number of hearths and a hollow. While a variety of activities were undertaken here, hideworking seems important, particularly in the vicinity of the hollow. This area is also adjacent to the channel and Brown and colleagues (in press) suggest this may have been an area for the manufacture and repair of small boats made from hide. A final phase of Mesolithic activity at the site dates to the 45th and 44th centuries. This consists of the building of a circular structure (structure 6), its position defined by knapping debris. Knapping sequences are more complete in this area, indicative of a late phase of occupation that underwent less subsequent clearance. Probably contemporary with this structure is a cooking pit, hearth and spread of burnt spread. These are associated with a dense knapping spread, which included the production of microliths in Southern Uplands chert. Tasks here were focused on the working of bone, antler and wood, with some butchery and hide working. A large number of microliths show impact damage, indicating repair of projectiles. 24 Lithic sources provide an understanding of the broader reach of the people who camped here. Eighty percent of the material used was beach flint and cherts that could be obtained over the course of a seasonal round that ranged from the Solway Firth to the upper reaches of the Eden Valley. More distant materials were also represented: pitchstone from Arran, Southern Uplands chert from southern Scotland, northern Pennine cherts, tuff from the southern Lake District and even flint from the chalk of the Yorkshire or Lincolnshire Wolds. Brown and Dickson (in press) suggest that the Solway Firth area may have been a place where groups from more distant regions met up. Scottish lithic materials may have been exchanged with groups whose range extended up the west coast or into the interior of the Southern Uplands; other groups may have moved south up the river Caldew and into the Lake District. Chalk flint from Yorkshire may have been obtained through exchange with groups from the east, with people meeting in the Pennine uplands. There may also be evidence for contact across the Irish Sea. Broken polished axes and reworked axe fragments made from Lake District tuff, including Langdale sources, have been recovered from Mesolithic contexts on the sites. One complete example is similar in form to polished axes common in the Irish Mesolithic. Ground stone axes are also known from west Wales (for example at The Nab Head II, see chapter 5), though these are rather different in form. Further south in Lancaster evidence for fifth millennium occupation has been found at Slynedales Culvert, on a slope above the valley of the Howsgill Brook and next to a small palaeochannel (Bradley and Howard-Davies 2018). Here a series of amorphous, possibly natural, features were sealed by a localised soil horizon [3087] containing Mesolithic lithic scatters; [3087] in turn was covered by a layer of colluvium. Mesolithic material was associated with the group of amorphous features sealed by [3087]. Small quantities of lithic material were recovered from two of these four features and all contained small amounts of burnt material, including in one pit hazelnut shells. One of these features has a radiocarbon date of 4650-4460 cal BC (5703±30BP; SUERC-68590). In the overlying layer [3087] three discrete Mesolithic lithic scatters were located, comprising in total 1475 pieces. The northernmost of these three scatters is the largest and associated with a cluster of burnt flint that may represent a hearth. The discard of microliths took place in the southern part of this scatter, but there is little evidence for their manufacture, and overall across the site numbers of tools are low. A flake from a group VI axe was recovered from the southwestern scatter, associated with a lithic assemblage that otherwise appears Mesolithic (ibid. 36) This may reinforce the evidence from Stainton West for the use of this source during the Mesolithic, though it should be noted that Neolithic features are present on the northern margins of this scatter. Further south still at Formby, footprints in contexts dating to the mid and late fifth millennium indicate people continued to move through the intertidal mudflats. Both adults and children, including very small ones, are represented at Blundell Path C (Burns 2021). Children seem to have been actively involved in tasks, often accompanied with an adolescent or young adult, perhaps following tracks, checking traps, fishing, fowling or playing. Burns’ studies paint a vivid picture of life on the saltmarsh, of paths and routeways alternately traversed by people and animals, red deer, roe deer and pig, and where cranes and oyster catchers alighted. The animals seem to have visited the area early in the morning, people later in the day, often walking towards the sea. <insert figure 6.16 here> Figure 6.16. Overlapping trails of human and animal footprints at Blundell Path A, Formby. Copyright Alison Burns. 25 The end of the Mesolithic in the uplands of northern England: The Pennines and North York Moors From the end of the sixth millennium BC, radiocarbon dated sites in the central Pennines become more prominent (figure 6.17), and it seems likely that this area played a more important role in peoples’ lives, shifting from an area of small temporary camps, to one that was at times more residential, though perhaps still only on a seasonal basis. The clustering of fifth millennium sites on long established routeways between east and west – such as the March Hill-White Hill-Windy Hill route (Preston 2012) – suggests these continued to be used, or at least places established through this historic use continued to be maintained. There may also have been more occupation on the North York Moors, though Mesolithic settlement is poorly dated in this region. In the final millennium of the Mesolithic dated evidence for human activity in the uplands comes from two sources: firstly evidence from occupation sites, particularly from the stone-built hearths and earth ovens that seem to be particularly prevalent during this period, and secondly from zones of high concentration of charcoal particles within environmental profiles, interpreted in many instances as evidence for human management of the landscape through fire. <insert figure 6.17 here> Figure 6.17. Pennine fifth millennium Mesolithic sites In both the Pennines and North York Moors evidence for burning of the landscape becomes prominent. This is partly due to the timing of peat development which makes identification and dating of these events possible; however data from North Gill on the North York Moors which has longer sequences than many other sites shows a greater frequency of firing events in the fifth millennium (Simmons 1996, 99). In the Pennines, peat began to develop in the deepest basins and around spring heads in the mid-seventh millennium, with a second phase of development from the mid-fifth millennium (Tallis 1975), though recent work reveals peat development in many areas was of later date (Garton 2017). The fifth millennium phase is perhaps broadly co-incident with the clustering of burning events at around 5600 cal BC noted by Albert and colleagues (2020). In certain areas fifth millennium firing practices seem to have been sufficiently persistent to trigger long-term landscape changes (Albert et al. 2020). In the Mid-Holocene, the tree line in the Pennines was probably around 500m asl, and tree stumps, preserved in peat have been found on some of the highest peaks. In the fifth millennium the central Pennines seem to have been characterised by a variety of vegetation types. Much of the higher upland areas were oak and alder woodland with an understory of hazel and some areas of open grassy heath (Tallis and Switsur 1990, Tallis 1991). In other places elm seems more important and lime and birch were also present (Bartley 1975), with perhaps more trees clustering in the more sheltered cloughs. In the Final Mesolithic Rishworth Moor, an upland plateau at c420m asl was covered with hazel and alder scrub with areas of grass and heather heath, while sedge fringed basins of marshy or open water. Trees (oak, elm and lime) probably grew on the plateau edge. The area saw three major episodes of firing between 4700 and 4100 cal BC. This seems to have been targeted at rejuvenating calluna scrub which had itself been established and maintained through burning practices (Albert et al. 2020). The longevity of burning practices may have been sufficient here and elsewhere in the Pennines to effect a long-term transformation of the landscape, as well as encouraging peat development. The highest point on the North York Moors, at Urra Moor, is 454m asl, meaning that deciduous woodland probably extended across the entire upland plateau in the late 26 Mesolithic. Pollen records indicate woodlands of pine and birch with oak and elm, with alder becoming locally important from the mid-fifth millennium (Simmons 1996). On the Moors too records of burning suggest extensive Final Mesolithic activity in the area, with dated fifth millennium burning events recorded at North Gill, Bonfield Gill Head, Bluewath Beck Head and Botany Bay (Innes et al 2012). Textbox 6.5: Mesolithic ‘firestick farmers’ From the 1970s onwards, palaeoenvironmental researchers began to note the persistent presence of micro-charcoal in mid-Holocene pollen profiles (Tallis 1970). While the causes of this charcoal can only be inferred rather than demonstrated, early explanations noted correlation between charcoal and reduction of tree cover and related this to records of ‘firestick farming’ (human management of the landscape through burning) in the ethno- historical record (Simmons 1975, Mellars 1976b, Jacobi et al. 1976). Other explanations for these patterns are certainly possible and Brown (1997) points out that charcoal accumulation and pollen profiles showing clearance and regeneration can be generated by a number of alternative factors. Charcoal from adjacent campfires is one possibility, given the frequency of Mesolithic occupation in many of these areas (Bennet et al. 1990). Tipping (1996) has highlighted the role of climate, suggesting that a synchronous rise in burning events around 7000 cal BC in Scotland is related to a shift to drier conditions rather than human action. Natural fires are likely to have played a role: pine burns more readily than other native arboreal species and was an important component of upland forests (Brown 1997, 135), though by contrast the damper Atlantic forest were less likely to burn (Simmons 1996). Lightning strikes, the main cause of natural fires, seem to have been more common at this time (Brown 1997, 135). Brown also notes the variable relation of fire to episodes of reduction in arboreal pollen: some of these episodes are not associated with charcoal, while others show charcoal associated with post-clearance profiles, possibly indicating fire at times played a role in preventing natural clearings caused by tree-fall from regenerating. Burning of woodland had been noted to increase the availability of plant foods such as hazelnuts which were exploited by humans; however much of the early emphasis was on the increased production of browse that might attract animals to predictable locations in the landscape as well as increasing their number and condition (Mellars 1976b). The presence of fungal spores deriving from animal dung immediately following burning episodes supports the idea that herbivores were drawn to these areas (Innes and Blackford 2003, Ryan and Blackford 2009). These practices, it has been suggested, could have led to incipient domestication (Jacobi et al 1976, 317) and potential corrals have even been suggested from evidence for clearance at Soyland Moor (Williams 1985). Burning practices have also been linked to the production of clearances for small-scale garden agriculture, based on the presence of cereal-type pollen in these sequences (Albert et al. 2020, Blackford and Innes 2020). Cereal-type pollen have been dated from c.4600 cal BC at sites such as.Black Heath (Ryan and Blackford 2009) and Soyland Moor (Williams 1985). However the pollen of cultivated cereals and some native grasses are near-identical in morphology leading to a scepticism over claims for Mesolithic cultivation. Recent work has suggested that the pollen of a primitive form of barley (Hordeum vulgare) can in fact be distinguished from the two native wild grasses with overlapping pollen morphology (Glyceria maxima and Glyceria fluitans) (Blackford and Innes 2020). On this basis Blackford and Innes identify the presence of cultivated wild barley in sequences dating to the last two of three centuries of the fifth millennium at Dog Hill, Rishworth Moor in the central Pennines. They also suggest barley is present in sequences from Cat Stones, also on Rishworth Moor, as 27 well as at Esklets on the North York Moors. Barley may have been cultivated rather than wheat as these are marginal areas for cultivation. Barley’s role in the production of alcoholic drinks may also have been an attraction (ibid. 378), particularly if the Pennines was a place where widely dispersed groups met. Pollen sequences following burning show a typical sequence of an initial phase of melampyrum and ruderal weeds, followed by a phase of heather, hazel and willow scrub, followed by woodland regeneration (Innes and Blackford 2003). Burning seems to have been used at times for different purposes. Early Mesolithic burning seems focused on wetland contexts (Day 1998, Chisham 2004), while Boreal and Atlantic burning became focused on woodland (or the prevention of woodland development), often seemingly focused on its margins (Bush 1988, Simmons 1996). This would encourage browse for animals and was perhaps also focused on increasing hazel (which is relatively fire-resistant) at the expense of other arboreal elements (Smith 1970). At North Gill on the North York Moors, some burning phases increased hazel, others open patches of grassland (Simmons 1996). In the Final Mesolithic of the Pennines, there are examples of burning used to regenerate open heathland. Burning also creates social spaces, small openings in the canopy that were used by people, where hearths could be built and earth ovens dug. The temporality of these openings may explain the clustering of features with a similar date at March Hill, and relate to the timing of woodland regeneration. Evidence from areas where more extensive surveys have been made, such as North Gill on the North York Moors, where fire cleared places along a stream, suggests areas of disturbance were in the region of tens of metres across (Simmons 1996, 86). Burning also signals the presence of people in the landscape to distant groups, useful perhaps is these were places where different groups aggregated. Nor should we ignore the symbolic significance of woodland and particular tree species and the impact of these broader ideas on the manipulation of woodland (Moore 2003). At Black Heath in the central Pennines the evidence suggests a natural opening in a woodland of hazel, alder and birch attracted animals, but that this clearing was subsequently maintained by people through burning over the following c250 years (Ryan and Blackford 2009). Peaks in charcoal suggest local firing occurred every 18-36 years, a slightly longer interval than the figures for optimal productivity of 5-15 years cited by Jacobi and colleagues (1976). Other areas show a variety of temporalities of firing events: at North Gill a series of sampling points close to a spring show both shorter (c10 years) and longer intermittent firing intervals lasting hundreds of years, though these latter may represent composite incidences (Simmons 1996). The microliths accompanying dated Pennine sites fall into two groups: Those dating to the first half of the fifth millennium are characterised by left lateralised scalenes and narrow backed bladelets, made in flint and chert. The second assemblage type contains only narrow, steeply backed bladelets (known as rods) made only in flint and dateing to the last few centuries of the Mesolithic (Switsur and Jacobi 1979, Griffiths 2012). Four-sided microliths and micro-tranchets may also belong to this period but are poorly dated in Northern England. There has been an inconsistent approach to the identification of four- sided pieces (see discussion in Preston 2012), but using Roger Jacobi’s identifications for consistency (Jacobi and Wessex Archaeology 2014), assemblages with four-sided pieces extend across the Pennines, but are rarer on the North York Moors (figure 6.2a). 60% of four-sided pieces are made on Wolds flint or similar white flint sources, the remainder seem to made on east coast till, with only two examples made on chert. This seems to indicate 28 eastern connections and may be a sign of groups from different regions aggregating in the Pennines. Microtranchets, both symmetric and asymmetric varieties, are more common on the North York Moors. They may be dated by an early fifth millennium measurement on oak charcoal from White Gill, though this site is complex (Carter 2014). This date would though be compatible with those from southern England for this microlith type. There is a near absence of dated sites on the North York Moors, though work by Spencer Carter is beginning to change this picture, with fifth millennium dates not just at White Gill but also at Esklets (Carter 2016). Beyond dated sites, microtranchets and other novel forms can perhaps provide potential clues to life in the region at the end of the Mesolithic (see figure 6.2). At Esklets, late Mesolithic occupation clustered around two spring-fed pools in a woodland of alder, oak, elm and hazel (Albert and Innes 2010). Site ECW1a was undated but composed of three small scatters, of which scatter B has two microtranchets, amongst nine microliths, retouched blades, an awl, a scraper and a burin. Thirteen cores may represent a cache (Carter 2016). Activity leading to the discard of a single micro-denticulate at Botany Bay, East Bilsdale Moor, also belongs to the fifth millennium (5956 ±51BP; Wk- 15138) through a date on the peat that encased it (Innes et al 2012). This is an indicator of off-site activity, perhaps related to harvesting or processing of the plant material that grew on this boggy ground. By contrast on the Pennines there are a number of fifth millennium radiocarbon dates for archaeological sites. Particularly notable are the cluster of dated features on March Hill, Dan Clough and Lominot. These sites are all in close proximity. There are around 120m between March Hill Top and March Hill Carr, around 250m between March Hill Carr and Dan Clough and around 400m between Dan Clough and Lominot. All these sites are dated to the fifth millennium BC. Stonehouse (1987) has previously highlighted the similarity of lithic assemblages from sites in close proximity in this region. He notes for example the similarity of the microliths (medium sized scalenes) and raw material use (around 20% black chert) on seven adjacent sites at March Hill South. Similarly, three sites with micro-scalenes have been found on the south site of Dan Clough, and ‘pears’, a rare type, are found on several sites on Dean Clough. These he suggests may represent broadly contemporary occupations. This is echoed by the patterning of the radiocarbon dated sites at March Hill Carr which suggest a number of hearths either used during a single visit, or more likely representing a history of repeated returns to the same place by a single group over a number of years. Given that many of these areas lay below the natural tree-line in this region, this probably represents activity in clearings and areas of woodland clearance that were maintained for decades and served as foci for repeated visits. The construction of stone-built hearths and earth ovens, some of which have evidence for repeated re-use, served as fixed points in these locations, with more temporary structures probably erected in relation to these. Modelling of radiocarbon dates from Black Heath indicates that the clearing was maintained for around 250 years with refiring roughly once a generation (Ryan and Blackford 2009). This may give some clue to the longevity of these upland site complexes and indicates concerns with maintaining places over many generations. The earliest dates from the March Hill area come from March Hill Mesolithic Project excavations on Lominot in 1996 (Spikins 2002). Three dates from a stakehole provide termini post quos for its infilling of 4990–4740 cal BC (Griffiths 2011, 155). Both early and late Mesolithic archaeology was found in this trench. The late Mesolithic assemblage is fairly distinctive, made on flint only, and showing use of the bipolar anvil technique, which is not common in the Pennines. Only three microliths were recovered, all scalene micro-triangles. A second date on bulked, unidentified charcoal of 4765-4235 cal BC (Q-1189) is available for Lominot IV, a site excavated by Francis Buckley; its exact location is unknown. An almost identical measurement of 4770-4455 cal BC (GrN-12278) (Stonehouse 2001) has been obtained on oak charcoal from a small stone-lined hearth excavated by Stephen Poole at 29 Dan Clough, located between Lominot and March Hill. This excavation, as well as an extension of this trench dug by the West Yorkshire Mesolithic Project, found an assemblage of 16 almost identical small scalene triangle microliths as well as other tools including several truncations (Conneller 1999). The homogeneity of the microliths probably argues in favour of a single occupation. The greatest intensity of settlement in this landscape is at March Hill Carr, a flat plateau at 430m asl on the southern flank of March Hill (figure 6.18). This is an area with a long history of excavation. Petch (1924) recorded four major ‘workshop sites’. Most relevant is Buckley’s hearth site, which appears to have been located in this area. Here two radiocarbon dates of 4930-4520 cal BC (Q-788) and 5490-4540 cal BC (Q-1188) come from one or two cooking pits (Buckley’s notebooks are ambiguous) around 30cm deep (Switsur and Jacobi 1975). Since Buckley’s excavation, March Hill has continued to be a magnet to both archaeologists and collectors. Local archaeologist J.L. Turner recorded in his notebooks in 1964: ‘March Hill, the mecca of all true flint addicts….is in absolute turmoil being slashed, hacked and torn to pieces in a most sacrilegious way’ (cited in Spikins 2002). Pat Stonehouse recorded at least eight sites on March Hill Carr during his work in the area (Griffiths 2011). <insert figure 6.18 here> Figure 6.18. View from March Hill, looking towards Pule Hill Between 1993 and 1995 the March Hill Mesolithic project undertook excavations at March Hill Carr (trench A) (Spikins 2002). This may be adjacent to Buckley’s site where one or two cooking pits were located (the adjacent area has been previously excavated and there were marks of an entrenching tool). In trench A, which measured only 4 by 5 metres, four features were located (figure 6.19). All have dates in the 48th and 47th centuries cal BC (Griffiths 2011) and all (and indeed Buckley’s hearth) could have been in use at the same time. Alternatively, the place could have been reoccupied over several decades. Of the four features, two were stone-built hearths, and one a cooking pit or earth oven, measuring 50cm wide and 20cm deep (Spikins 2002). This was filled with charcoal and burnt stones. Black and Thomas (2014) argue that on upland plateaux (such as March Hill Carr), earth would have remained visible for centuries, and thus encourage reoccupation and the emergence of a ‘persistent place’ (Schlanger 1992). The visible presence of this facility may account for the quantity of features, with similar radiocarbon dates, that surround it. <insert figure 6.19 here> Figure 6.19. Trench A and associated features The final feature at Trench A was a small pit crammed with burnt, worked flint and charcoal. The flint was white and heavily fragmented. While this may represent an exercise in heat- treatment that went wrong (heating improves flint’s flaking quality), there is no evidence that any of the remainder of the lithic assemblage has been heat treated (Conneller 1999). This may represent a more formal act of deposition with lithic material associated with a particular event intentionally destroyed. Burning flint transforms it, making it appear similar to burnt bone, perhaps representing death. More evidence for depositional practices can be found in material in the earth oven, which has not been left with the basal layer of charcoal and stones intact; instead it seems to have been reused for midden material, including charcoal and burnt stones and worked flint. Charcoal within all four hearths consisted of oak and hazel, likely to be indicative of the local vegetation of the uplands at the time. More rarely represented in the hearths are hawthorn, Sorbus sp. (probably rowan) and Prunus sp. (probably blackthorn). All but one of the five hearths excavated during the project contained either hawthorn or blackthorn, both useful kindling species (Spikins 2002). 30 March Hill trench A No. Burin 7 Burin/scraper 1 Borer 1 Microlith 91 Notch 1 Scraper 3 Retouched/utilised 167 Burin spall 9 Microburin 7 Core preparation 57 Core 45 Nodule 1 Blade 269 Debitage 1729 Total 2381 The lithic assemblage indicates that a wide variety of activities were undertaken with large numbers of retouched and utilised pieces. Formal tools are dominated by microliths: Of 62 complete microliths at Trench A, 51 were triangles. March Hill Carr, not unexpectedly, thus fits well into Jacobi’s ‘March Hill type’ site (Switsur and Jacobi 1975), characterised by small, narrow scalene triangles with lower numbers of other geometrics, mostly backed bladelets. Microburins are relatively rare but more common in a small trench (73) a few metres to the east, possibly indicating different activity zones. Small neat truncation burins are another feature of the assemblage and March Hill has long been known for the quantity of burins recovered – 29 for example from March Hill 2 (Buckley 1921, Petch 1924, Stonehouse 1987). These are less common elsewhere in the central Pennines, though several are known from Dean Clough G and Badger Slacks 2 (Stonehouse 1987). Burins were common elements in Star Carr type assemblages, but decreased in importance in Deepcar and particularly Middle Mesolithic toolkits. This renewed interest in burin manufacture might be related to the renewed production of bone and antler tools. With the exception of bevel- ended tools and an Obanian style uniserial barbed point from Druimvargie, these are not common between c.8000 cal BC (when Star Carr style uniserial barbed points fall out of favour) and c.5500 cal BC. After this time both T-shaped antler axes and biserial barbed points are relatively common (Tolan-Smith and Bonsall 1999, Elliott 2012). While burins are not needed to work antler (a snapped blade will work just as well), an increased role for antler tools may have led to the reintroduction of a more formal toolkit for their production. Much of the lithic distribution at March Hill Carr appears to focus on the north-east side of the hearths. The highest lithic densities are here as well as concentrations of microliths, retouched and utilized pieces, blades, core maintenance flakes and chips. These indicate that activities involving routine core reduction, the replacing of elements of composite tools and cutting activities centred round the four hearths, possibly allowing for prevailing wind direction. The cutting activities may represent food preparation, as these are found next to the earth oven. Moderate lithic densities are represented – up to 275 pieces per square metre. However this is a landscape where lithic material is not readily available and some curation is likely. 31 The variability in Pennine Mesolithic assemblages usually characterised as microlith dominated hunting camps has been discussed in chapter 5. Sites from the fifth millennium show a similar dominance of microliths, but apart from Rocher Moss South 2, all have numbers of other tools, and given the uncertainties around microlith function and impact of resharpening practices the general picture is of very varied activities (table 6.9). There is perhaps less evidence for more specialised activities than the Pennine sites discussed in the previous chapter, but there is some variation: burins are frequent, at March Hill Carr, and South Haw, while scrapers are well represented at Dunford Bridge B. The evidence for varied activities, taken in conjunction with the density of stone-built hearths and cooking pits at sites such as March Hill Carr reinforces the suggestion that Pennine Final Mesolithic sites are something more than simple hunting camps. Site Microliths Scrapers Burins Awls Truncation Scalene triangle sites Dan Clough 2 15 2 1 1 3 March Hill Carr 91 3 8 1 0 Rod sites Dunford Bridge B 19 6 2 0 2 March Hill Top 15 0 1 0 0 Rocher Moss 1 30 2 1 0 0 Rocher Moss South 2 35 0 0 0 0 South Haw RA1 42 2 5 1 0 South HawTP1 23 2 4 0 0 Table 6.9. Essential tool frequencies for Fifth Millennium Pennine sites Late/Final Mesolithic sites have been traditionally considered as being smaller in size than Early Mesolithic examples. Both Petch and Buckley noted that ‘there is seldom found an indefinite scattering of flint over a wide area’ (Petch 1924, Buckley 1921, 1924), an observation confirmed by Mellars (1976). While the Late Mesolithic scatter on Lominot appears small and localised, the March Hill Carr site appears rather different. March Hill Carr, though damaged by collectors and early excavations, could once have extended to 214m² in area. Earlier digging activities make it difficult to discern the patterning of this activity beyond trench A but the recovery of one or two cooking pits at Buckley’s March Hill 2 site suggests the density of features continued. The series of hearths and the similarity of the radiocarbon dates from two of these features suggests that March Hill Carr saw fairly intensive use over a short period of time, perhaps a regular camping site, perhaps, given the presence of hazel wood but lack of hazelnuts recovered, one with a seasonal basis. A summer aggregation site is a possibility, and the presence of several cooking pits spread over a small area, might indicate a place of feasts and celebrations. Large spreads of late Mesolithic material have also been noted on the North York Moors, often around springheads, for example at Snilesworth Moor and Upleatham (Simmons 1996), though as these are undated it is unclear whether or not they represent a few, temporally clustered larger encampments or repeated, small-scale visits over millennia. Recent surveys have shown concentrations of small late Mesolithic scatters in particular areas: Bransdale moor, the Farndale-Westerdale-Baysdale watershed, Glaisdale-Rosedale moors, West Bilsdale and Snilesworth moors (Waughman 2017). These may represent repeated return to particular places on a similar temporal scale to the Pennine sites. Despite investment in cooking pits and stone-built hearths, evidence for additional structures amongst fifth millennium sites is slight. A single stake hole was noted in the very small trench at Lominot C (Spikins 2002). In the northern Pennines two features have been found at the two rod sites at South Haw (Chatterton 2006). The most extensive was in TP1 where 32 an earth oven was found surrounded by an arc of stakeholes enclosing an area measuring around 2m in diameter. A larger posthole was located adjacent to the firepit which was surrounded by dark sediment. The pit seems to have been reused and filled with debris from the occupation. At RA1 a shallow scoop measuring 2 by 1.4m was found which seemed to represent an occupation floor, though without associated posts or stakeholes. The last hunters in the uplands Little has thus far been said about the final site excavated by the West Yorkshire Mesolithic Project. This is March Hil Top (trench B), located just below the brow of the hill. Here a small scatter of lithic material was recovered, surrounding a stone-built hearth that had two phases of use. Material cleared out from the hearth seems to have been dumped adjacent to it, rather than placed within a pit, as was often the custom at earlier sites. This evidence from the top of March Hill differs from that on March Hill Carr: The scatter is very small in size, most lithic material derives from only two square metres; it is made on translucent flint only, rather than the mix of flint and chert at trench A, and the only microliths represented are straight-backed bladelets, often known as ‘rods’. About 300 of these ‘rod’ sites are known in the north, and are confined to the uplands of the Pennines and the North York Moors, though stray finds, probably representing composite tools, are known from the lowlands of northern England (Jacobi 1978, Switsur and Jacobi 1976). As Griffiths (2014) notes, the term ‘rod’ has been used very variably in Mesolithic studies. Jacobi used the term rod to describe any narrow backed bladelet where the backing seemed designed to produce a straight edge; it was originally employed by Rankine to refer to a double-backed bladelet. However others have used rod to mean any form of backed bladelet microlith, while others reserve the term to refer to steeply retouched single or double backed bladelets (the latter sometimes more akin to Jacobi’s needle point; Preston 2012) which perhaps appear more characteristic of the Pennine final Mesolithic sites. While traditionally associated with the very end of the Mesolithic, narrow straight-backed bladelets occur throughout assemblages traditionally defined as late Mesolithic. They are present in the earliest of these, such as Filpoke Beacon, where they slightly outnumber scalenes, and in much later ones, such as the early fifth millennium BC assemblage from March Hill Carr, where scalenes are more dominant (Conneller 1999). A find of a composite tool composed only of backed bladelets from Seamer K in the Vale of Pickering has been radiocarbon dated to 7570-6805 cal BC (HAR-6498) (David 1998). Ellaby (2987) also makes a strong case that straight-backed bladelets were the dominant microlith type in southeastern England around the same time. However there are also a series of dated assemblages with steeply retouched straight backed bladelets as the only microlith form present that date to the very end of the Mesolithic, at least in the Pennines where they are associated with small sites at high altitudes and made entirely or predominantly on flint. The combination of these specifically rod-like microlith types, raw material differences, site size and upland location in the Pennines does provide greater confidence in assigning undated sites with these features to this distinct chronological phenomenon. Several radiocarbon dates from the March Hill Top hearth indicate that it was in use between 4190-3970 cal BC (Griffiths 2011). Three other of these ‘rod’ sites have dates. From the central Pennines are Dunford Bridge B at 4360-4000 cal BC (Q-799, 5380±80BP) and Rocher Moss South 2 at 4935-4460 (Q-1190, 5830±100BP). Both are on unidentified bulked wood so should be regarded as termini post quos for rod assemblages, while the date on charcoal from Rocher Moss South 2 does not come from a sealed context and thus may be particularly suspect (Jacobi 1993). Finally in the Northern Pennines, South Haw in 33 Nidderdale has radiocarbon dates from two separate hearths, both on birch charcoal, indicating occupation between 4230-3985 cal BC (Beta-189653, 5270±40BP) and 3945- 3705 cal BC (Beta-189652, 5010±40BP) (Chatterton 2006, Griffiths 2014). These indicate rod sites in this region date to the very end of the Mesolithic, at a time when people pursuing Neolithic ways of life were already present in Britain (Griffiths ibid.). March Hill Top is microlith dominated (see figure 1.5), though a burin and resharpening spalls are also present. Other rod sites show more variability (table 6.9). At Dunford Bridge B, though microliths were most common, a relatively high frequency of scrapers and two burins were recorded. South Haw has 42 microliths but six burins, two scrapers and an awl (Chatterton 2006), while the small assemblage of Rocher Moss South 2 yielded only microliths (Stonehouse 1980). The March Hill Top scatter is small, only 9m² in area. Rod sites in the Pennines tend to be small in size, with assemblages of less than 650 pieces; larger sites are present on the North York Moors thought these are undated (Jacobi 1975, Switsur and Jacobi 1976). Jacobi (1975) noted they tended to be found on the higher points of the Pennines, a finding recently confirmed by Preston (2012) who notes they are confined to the top 20% of the central Pennine altitudinal range. The March Hill Top site is similar to Rocher Moss South 2, in its small area, dominance of microliths and high frequency of burnt material, which is likely in both cases to be the result of site/hearth maintenance activities. A feature of many rod sites is a focus on the production of relatively (for the Pennines) large blades. This has been noted at South Haw (Chatterton 2006), Rocher Moss South 2 (Stonehouse 1972), and in a group of rods accompanied by a group of blades measuring around 40 by 12mm from Tintwistle site MST01 ACC, which may represent a cache (Garton 2017). Chert is rare or absent at rod sites, instead the dominant material is flint: only five pieces were recovered at March Hill Top (Conneller 1999); of the 42 rod microliths at South Haw RA1, only one was on chert, and all 25 from TP1 were on flint (Chatterton 2006). A translucent brown flint was used at March Hill Top and South Haw TP1, a grey opaque flint at Dunford Bridge B and a mottled grey flint at South Haw RA1. Both March Hill Top and the two rod sites at South Haw have low quantities of cortical flakes in comparison to the triangle dominated assemblages, indicating flint cores were brought in already partly reduced (Conneller 1999, Chatterton 2006). This, combined with the evidence for the production of large blades, may suggest changes in technological and transportation strategies in comparison to the triangle sites where Myers (1986) has argued that both flint and chert were brought in as unworked nodules. These changes may indicate shifts in mobility or may be the product of the emergence of new networks of exchange and obligation which guaranteed previous supplies. There is evidence for disturbances in the local vegetation in the last centuries of the fifth millennium suggesting that burning of the vegetation continued as a practice amongst groups using rod microliths. This can be seen at Soyland Moor, for example, where late burning is accompanied by the presence of cereal-type pollen dating to the 42 nd or 41st centuries BC (Williams 1985). There is also very late Mesolithic evidence for burning at Rishworth Moor, where a hiatus in the charcoal record of around a century perhaps marks the end of the moor’s use by Mesolithic groups before the area was reoccupied during the Neolithic (Albert et al 2020, 13). Similar evidence is present on the North York Moors where burning episodes dated to between the 42nd and 40th century cal BC have been recorded at Esklets, Bonfield Gill Head, North Gill and Bluewath Beck Head (Albert and Innes 2015, Innes et al. 2010). There are suggestions that small scale disturbance episodes at Esklets were associated with the 34 cultivation of barley, and cereal-type pollen has also been found associated with a sequence of this date at North Gill 1A (Simmons and Innes 1996). A TPQ for oak charcoal of 4360– 4250 calBC (5449±30 BP; SUERC-62297) from Esklets ECW2 is associated with a shallow cooking pit and a small knapping scatter focused on the working of small, partially worked cores and scalene microtriangles (Carter 2016).There have been suggestions that a rod microlith phase also existed on the North York Moors (Spratt et al. 1976), though this has not been confirmed by radiocarbon dating, and inconsistency in the use of the term ‘rods’ and the persistent presence of backed bladelets throughout the late Mesolithic mean this phase cannot yet be substantiated. Two sites at Esklets have been suggested to represent rod sites, site ECW6 and Hayes’ Esklets site. These are both small assemblages whose only microliths are straight backed bladelets (ibid). North Wales Fifth millennium radiocarbon dates are relatively rare in north Wales, consisting often of determinations from isolated pits, as at Rough Close, Powys and Parc Bryn Cegin, Gwynedd. The only area that has seen more systematic investigation is Nant Hall, Prestatyn. Here several small shell middens have been located, lying on the eastern side of what was then the estuary of the river Cwyd. To the south the land rises steeply to the limestone of the Clwydian hills. At a time of higher sea-level, an estuary around 9km wide was created, stretching from Abergele to Rhyl. Excavations at Nant Hall were carried out in 1991 by CPAT in advance of development and identified three middens, with additional fieldwork carried out the following year in conjunction with University of Wales, Lampeter, during which two further middens were located (figure 6.20). Two middens, composed primarily of cockles, were dated to the early Neolithic, and two others (D and E), composed of mussels, were dated to the last few centuries of the Mesolithic (Bell 2007). This change in composition may reflect changing coastal morphology as a result of shifting sea-levels. Mussels are species of the rocky shore, while cockles are associated with sandy and muddy sediments. <insert figure 6.20 here> Figure 6.20. Excavations at Nant Hall, Prestatyn. Copyright Martin Bell Environmental evidence indicates that the vegetation near the mussel middens was woodland, possibly with small clearings. The middens themselves are close to the wetland edge, probably in an area of mixed deciduous woodland of oak, hazel and lime. Saltmarsh lay beyond, with the spring tide highwater level around 45m distant (ibid, 309). The midden at site D was located on a palaeosol that itself contained charcoal and occasional worked flint which has been dated to 4470-4050 cal BC (5470±80BP; CAR- 1424). The midden was between 2 and 20cm thick. Its full extent was not uncovered, but it measured 5m from north to south and at least 5.5m from east to west. It appeared to have been composed of several merging, smaller heaps of mussels and other shells (periwinkles, cockles and oyster), a few animal bones (red deer) and occasional lithics and burnt stone. Beach pebble flint was predominantly used, a contrast to the reliance on chert at the earlier site of Prestatyn (Bryn Newydd) (see chapter 3). Charcoal from the midden itself dates to 4330-3950 cal BC (5270±80BP; CAR-1423). The midden at site E was smaller, only 2m by 3m and 10-50mm thick, and composed of mussels with occasional cockles, periwinkles and welks. Some animal bone was preserved 35 but only a single red deer antler was identifiable to species. Two dates, one on the lower part of the midden, the other on the upper part indicate use in the second half of the fifth millennium BC, perhaps into the first part of the fourth. Two additional middens (F and G), both composed of mussels, and thus likely to be Mesolithic, were located but not excavated. Augering suggests F was less than 5m in diameter, and G measured around 3m by 1.5m. The small size of the middens suggest these may have played a short-term, specialist role. The Midlands There are only three fifth millennium radiocarbon dates in the whole of the Midlands that might be reliably attributed to human activity, and as ever the nature of the Mesolithic in this region remains elusive. Two early fifth millennium dates on roe deer bones incorporated into the pre-barrow Neolithic midden may indicate final Mesolithic activity at Ascott-under- Wychwood. A handful of left lateralised scalene triangles and backed bladelets might be compatible with this date: Four scalenes were found beneath the barrow and two scalenes and two backed bladelets were recovered from the area of the Roman quarry around 10m distant from these, which are likely to have moved down slope (Jacobi archive, British Museum). However distinguishing the extent of final Mesolithic activity on a site disturbed by Neolithic activity within an assemblage including Early Mesolithic types is difficult. The remaining radiocarbon date comes from a wooden stake from a palaeochannel at Manor Farm, Milton Keynes, an area where there are late Mesolithic lithic scatters (Billington 2017). Microlith typology may provide a few more candidates if symmetric micro-tranchets, like their asymmetric cousins, also belong to the fifth millennium, and if a late date for four-sided pieces can be confirmed. Over-Whiteacre Spring in Warwickshire, a site with extensive Honey Hill and possibly also late Mesolithic occupation has yielded a handful of asymmetric micro-tranchets and four-sided types (Saville 1981a, figure 9). Four-sided microliths are also present in the Sturge collection (BM) from Lakenheath (Jacobi 1984, fig 4.15) indicating that this location – another noted long-ago by Clark as a node of Mesolithic activity – remained a focus at the end of the period. Four-sided pieces are also present at Peacock’s Farm, Shippea Hill (as are both asymmetric and symmetric micro-tranchets) (Clark et al. 1934), another place repeatedly revisited in the Mesolithic, as well as Lackford and Stow, Suffolk. Micro-tranchets are also known from several surface collections in the region. These have mainly an eastern distribution, with examples found at Risby Warren, Crosby Warren, Roxby, Salmonby, Manton Warren, West Keal and Belchford in Lincolnshire, New Fen at Lakenheath, Cavenham Heath, West Row Fen and Eriswell in Suffolk; but there are also finds in the west at Long Newnton and Syreford in Gloucestershire (Jacobi and Wessex Archaeology 2014). The Thames and its tributaries Fifth millennium evidence comes mainly from the west of the area, from the Colne and its tributaries and from the Kennet. Evidence for the use of the Lea is currently lacking but there are two sites with microtranchets further north on the Essex river, the Blackwater (Jacobi and Wessex Archaeology 2014). To the south asymmetric micro-tranchets are present at Orchard Hill, by the springs that form the head of the Wryse that flows into the Wandle, and a place repeatedly occupied in the Mesolithic. A stray antler beam mattock from Staines of fifth millennium date (Elliott 2012) could come from a Thames or Colne related context. On the Thames, evidence for continued, perhaps even intensified occupation, at Runnymede Bridge has been recorded, as a silt island formed which was less affected by flooding. 36 Residual charcoal recovered from a ditch falls between 4835 and 4450 cal BC (5780±85; OxA-3582). Three microliths, including a needle point and a scalene, were also recovered from the ditch. Five of the 25 microliths from the site are needle points. TL dates on two clusters of burnt flint probably also indicate late fifth millennium occupation (Needham 2000). There is some evidence for continued occupation on the Colne. In 1971, around a kilometre to the west of Tolpits Lane B101 (see chapter 5), the partial remains of an aurochs were located by the Rickmansworth Historical Society in peaty deposits overlying Colne Terrace gravels. These consisted of lumbar vertebrae, part of the frontal and a left horn core, the base of which revealed cut marks associated with skinning. This has a radiocarbon date that falls into the second half of the fifth millennium BC (4500-4260 cal BC at 1σ; 5540±110BP, BM-1676R) (Burleigh et al 1982, 263). The more significant fifth millennium sites lie in the upper reaches of the Colne’s tributaries, following the late Mesolithic pattern. Excavations at Misbourne Viaduct were small in scale, and undertaken in a short period prior to the building of the M25. This work yielded an important assemblage of lithics and animal bone beneath and within tufa. Microliths consisting of small scalenes and narrow backed bladelets were associated with three radiocarbon measurements on samples from the two lowest layers which date to the late sixth or early fifth millennium BC. An extremely varied faunal assemblage came from a small area including aurochs, red deer, roe deer, pig, beaver, wild cat, otter and badger. Wood charcoal from the site included oak and ash (Farley 2010). More detailed evidence is available from another small excavation, at Stratford’s Yard, Chesham. At Chesham several narrow valleys meet and the waters of three chalk springs converge to form the river Chess. Near to the confluence of the most northerly of these, the Vale Stream, with the Chess, a dense spread of Mesolithic debris was located. Overlying floodplain alluvium are what may represent a buried soil containing the majority of the artefacts. Overlying this was around 60cm of colluvium, containing material of a range of dates, from post-medieval to Mesolithic. Behind the site the ground rises steeply to the east and the chalk scarp of the Chilterns. As well as burying the site, erosion of the Upper Chalk down this slope would have provided flint nodules for the site’s inhabitants. The integrity of the possible buried soil is uncertain. Much of the later prehistoric material (Neolithic flintwork and Iron Age pottery) present on the site derives from the colluvium, as does a substantial quantity of medieval and post-medieval pottery. However the majority of the small assemblage of highly fragmented Iron age pottery comes from the uppermost 5cm (context 4/X) of the potential buried soil sequence. Charred cereal grains are also found throughout the sequence. These have been suggested to be medieval in date, though would also be compatible with the Iron Age evidence. It may be that the buried soil was active until the Iron Age, however perhaps a more likely scenario is that the context 4/X, thought to be in situ, is an older colluvial layer, burying the site. This uncertainty as to the nature of these sediments means that question marks remain over the reliability of a radiocarbon date on bulked bone from the site. Fieldwork was undertaken by the Chess Valley Archaeological and Historical Society (CVAHS) in 1969 and 1982, who excavated two small trenches in advance of development. The 1969 trench measured 4x3m; the 1982 test pit, located 2m away from the earlier excavation, only 1m2. Service trenches for the development were also monitored and lithic material recovered (Stainton 1989). The area to the north of the site also underwent excavation in 1989 when substantial quantities of lithic material were recovered. While this is of mixed date it includes an important Mesolithic component, including microlith types not 37 represented in the CVAHS excavations. In all, it appears that a large and complex site is present, with occupation spanning the Middle to Final Mesolithic. Examination of the microliths from the CVAHS excavations indicates that they fall broadly into two or perhaps three groups. The first consists of obliquely blunted points and part- backed points, curve-backed pieces, lanceolates and microliths with inverse retouch. These would be compatible with a Middle Mesolithic date and though lacking classic Honey Hill Points, are broadly comparable to assemblages from sites such as Asfordby (see chapter 3). This activity may be dated by two radiocarbon dates on burnt hazelnut shells from the lowest spits of the excavation that fall into the second quarter of the eighth millennium BC. The larger group consists of small scalenes with three sides retouched, small crescents, symmetric microtranchets and asymmetric microtranchets (figure 1.5). One of the latter has a lower concave truncation which forms a tang which is covered on the inverse with invasive retouch. These highly distinctive types have been named ‘Bexhill Points’ (Donnelly et al 2019). Microtranchets are diagnostic of the latest Mesolithic industries, and, along with four sided points, were noted by Jacobi to belong to the last millennia of the Mesolithic (Jacobi 1978, 19). A group of backed bladelets, most with a single retouched edge, may belong with them but could easily be earlier. A date of 5005-4505 cal BC (5890±100; BM-2404) obtained for the site (Stainton 1989) was perhaps significant in Jacobi’s suggestion of a late date for microtranchets and similar forms. However this date may be problematic: it is on bulked bone from the lower part of the possible buried soil (context XII/V) that may have been sealed rather later than the Mesolithic. In its favour the sample comes from a layer that has relatively little evidence for intrusive material and was made up from three bones from a single wild species (aurochs). The measurement is also compatible with new dates for similar lithic assemblages from Bexhill (Donnelly et al 2019). Though the small size of the CVAHS excavations precludes detailed understanding of the Mesolithic occupation at Stratford’s Yard, it is clear that this is a large and complex site that was repeatedly reoccupied over the course of the Mesolithic. Features were also present: four small pits or postholes were recorded. The tools recovered indicate a variety of activities: microlith production and retooling, tranchet axe use and resharpening and use of scrapers. A large assemblage was recovered, despite some material being lost. The new dating indicates processing of hazelnuts took place in the Middle Mesolithic, while processing of faunal remains – mostly aurochs, but also red deer, roe deer, pig, and possibly frog and birds – occurred, perhaps towards the end of the Mesolithic. Continuity of occupation has also been noted on the Kennet. The large Mesolithic site of Wawcott III has amongst its wide range of late Mesolithic types a small Final Mesolithic component of four-sided pieces and asymmetric microtrachets. A late radiocarbon date of 5345-4725 cal BC (6120±134; BM-767) comes from bulked material from pit 2 (Froom 1976, 160). This measurement, though not without its uncertainties, is compatible with other dates for microtranchets and 4-sided microliths. Pit 2 is a hollow that was only partially excavated as it extends into the section, though augering of the unexcavated area revealed it measured around 3.5mx1.5m. Its morphology suggests it may represent a tree throw, albeit one that may have seen human use, as its base contained concentrations of burnt stone, charcoal and charred hazelnut shells. A similar date has been obtained for late Mesolithic activity just 1200m to the east at Wawcott XXIII (Froom 2012). This site is on the floodplain (figure 6.21), close to a relict channel. Lithic material numbering just over 5000 pieces was recovered from the lower of two layers of organic sand-silt, sealed by a layer of marl or tufa. Particle size analysis suggests this sand may be windblown and that occupation took place on a low river dune, 38 capping a gravel rise (Barnett et al. 2019). While a small early or middle Mesolithic component may be present, the majority of this assemblage is late and final Mesolithic. Microliths consist of scalene triangles with three sides retouched, micro-crescents and a handful of double-backed bladelets (Froom 2012, figure 7.5). Several asymmetric petit tranchets indicate a final Mesolithic component and a date on charcoal from a small hearth of 5295-4730 cal BC (6079±113; BM-2404) would be compatible with this element. <insert figure 6.21 here> Figure 6.21. View across the floodplain at Wawcott. In addition to this hearth, spreads of charcoal and burnt stones were relatively common at the site. A single pit, 1m in diameter, was located; this was infilled with silt containing flint, burnt stone and charcoal. Some faunal remains were present, though poorly preserved, and include aurochs, red deer and pig (all of the latter possibly from the same individual) (Carter 1975, Froom 2012). A single human tooth, a lower right incisor from an elderly individual, was also recovered (Froom 2012, 238-9). It is not known whether this was lost ante-mortem, and none of the bone from the site is dated. The lithic assemblage is dominated by microliths and microburins (table 6.10). Scrapers are rare, with only two recovered. Burins and truncations are common, though some of the former appear to have served as bladelet cores. A large axe, possibly a roughout, was recovered, made from Dorset chert (Froom 2012). Tool/tool spall Wawcott I Wawcott XXIII Axe/pick 5 1 Axe flake 4 0 Burin 4 31 Burin spall 4 0 Microlith 51 58 Microburin 61 30 Micro-denticulate 1 0 Notch/denticulate 0 3 Scraper 5 2 Truncation 0 40 Retouch/utilised 55 present Cores 72 76 Table 6.10. Tools from final Mesolithic Kennet sites At Wawcott I, occupation was located on a low gravel spur. A 6.5x5.5m trench uncovered a sub-circular feature that may have originally been a tree throw, but subsequently underwent human modification. A pile of flint nodules from the underlying gravel close to the feature may suggest it was as a source of flint (gravel flint was the main source used at this site). Subsequently four post holes were dug either side of the feature (three on the east, one on the west) and a hearth built in the depression, marked by burnt flint and charcoal. One piece of charcoal gave a radiocarbon date of 4350-3790 cal BC (5260±130BP; BM-449) (Froom 1971). 39 The toolkit at the site is characterised by microlith production debris and discarded microliths and utilised flakes and blades. A wide variety of other tools are also represented, albeit in small numbers (table 6.10). Notable is the presence of seven axes, including an axe of black Portland chert recovered from initial fieldwalking of the site, and two resharpening flakes (Froom 1971). The microliths are dominated by four-sided pieces, suggesting most of the material belongs to a single occupation. If the four-sided microliths belong with this radiocarbon measurement, this would suggest a very late date for this type. South of the Rivers Typological evidence for this period indicates continued occupation of favoured places. At Farnham, symmetric and asymmetric micro-tranchets suggest that occupation continued c on a site that had been a persistent place for several millennia. Micro-tranchets are also present at both North Park Farm, Bletchingly and Puttenham Heath on the Lower Greensand, and at Roffey Holt, the well-known Horsham and late Mesolithic site in the Ashdown forest. There are also fifth millennium dates for the Wealden rockshelters, though as these are based on bulked charcoal from sites with Neolithic evidence they should be treated with some caution. Other sites with long histories in the region such as Selmeston or Beedings Wood show no visible evidence for fifth millennium activity. The richest evidence for fifth millennium settlement comes from Combe Haven and the excavations in advance of the Bexhill relief road. Here a large number of Final Mesolithic scatters are associated with Bexhill points and a range of other microliths (Champness et al. 2019). In common with the sites described above, Bexhill was a place that had been revisited for millennia, though the environment and geomorphology of the place changed considerably over this period. Following a period of intermittent marine incursions during the late Mesolithic when the area was less frequently visited, in the fifth millennium the landscape became more stable. Peat began to develop in the lower reaches of the valleys in a landscape of alder carr, dissected by freshwater streams. Higher, drier ground was characterised by mixed lime/hazel/oak woodland, while ferns grew in clearings and on the woodland edge. Mesolithic occupation is concentrated on drier spurs and peninsulas extending from the valley margins into the wetlands, with the highest densities found on site 15, a spur sloping down towards a river channel (ibid.). Of the 210 lithic scatters located by the excavations around 60% belong to the Late/Final Mesolithic. Of these, dating and typochronological work indicate that the majority belong to the period 5200-4000BC, with assemblages containing Bexhill points dating to between 5200 and 4200BC (Champness et al 2019). The area seems to have become more attractive as it became more stable and ceased to be subject to intermittent marine incursions, perhaps as a barrier developed, though the sea would still have been close. Most scatters are microlith dominated, though other formal tools were also found within most of them. Features associated with the scatters include hearths, stakeholes, pits and tree-throws used for the deposition of lithic material. Depositionary practices seem to have focused on the debris from specific events rather than the generation of middens that accumulated over a long period of time (ibid). The full publication of this site will undoubtedly transform our understanding of this period and provide a nuanced history of the last Mesolithic groups and the transformations leading up to and encompassing the appearance of elements of the Neolithic package. Beyond Bexhill, radiocarbon dates come mainly from pits. A late episode of pit-digging was located at Charlwood in Surrey, amongst what was an extensive lithic scatter, now truncated by the plough (Ellaby 2004). More than 21,000 lithics were recovered from the scatter, the 40 densest part of which was centred on the area where the pits had been dug. Seven pits were excavated which enclosed an area of 15 by 9m. All were heavily truncated. Pit one survived best, to 32cm depths and with a diameter of 2.24m. All contained flint: Quantities of lithics in each feature range from 22 to 1088, with an average of 398. Given that these pits are all heavily truncated, they must have originally contained substantial quantities of material which seem an unselective sample of on-site debris. The tools are dominated by microliths: Bexhill points (inversely retouched oblique microtranchets) and microscalenes. Burins and truncations are also represented; scrapers, as at the Wealden rockshelters, are rare. Microburins outnumber microliths indicating a focus on retooling. Burnt flint is common with up to 50% of the total flint in pit 3, and an average for all pits of 36%, though this is similar to that from the ploughsoil. Three pits contained burnt bone as well as flint. Larger pieces from pit one may represent roe deer. Three dates on bulked charcoal from the lowest fills of pit 1 returned inconsistent late Mesolithic and early Neolithic dates, which could indicate contamination or old wood effect. Given the lack of Neolithic evidence at the site, it is likely this pit belongs broadly to the fifth millennium BC. The Southwest There also appears continuity in the use of favoured locations in southwest England, with several of the key seventh and sixth millennium sites discussed in the previous chapter showing evidence for fifth millennium activity. Faunal remains have been recovered from several of these sites. Species diversity appears lessened, though the numbers involved are very small (table 6.11). At Langley’s Lane a final Mesolithic phase, thought to date from this period, followed the cessation of tufa deposition. At this time the tufa would have formed a low mound into which were dug a large pit, two postholes and six stakeholes. A layer of stones was spread across the edge of the tufa mound, where the white of the tufa and brown of adjacent sediments would have made a notable contrast. Within the layer of stones was a deposit of lithics and animal bones (Lewis et al 2019). Radiocarbon evidence indicates occupation also continued at Blashenwell, Site Aurochs Red deer Blashenwell P Blick Mead (dated fifth P P millennium bones only) Langley’s Lane 4 1e (5th millennium?) Table 6.11. Fauna from fifth millennium contexts in southwest England At Blick Mead there may, if anything, be a greater intensity of use in the fifth millennium. In the lower lying trench 19, which was dense with animal bone and where the focus of radiocarbon dating efforts were concentrated, slightly more Final Mesolithic dates were obtained than those belonging to previous millennia (Jacques et al. 2018; see table 6.12). There also appears evidence for Final Mesolithic activity on the dryland trench 24. Here the lower fills of a possibly modified tree throw have two late fifth millennium dates, albeit on oak. There is also a late fifth millennium date on unidentified charcoal from a possible post- hole a few metres to the north, and a fourth date on oak charcoal from a layer described as loess within the same trench. An lithic assemblage of 808 pieces, including 20 microliths, two truncations, a notch and a piercer as well as several utilised flakes and blades, was recovered from the tree throw. The microliths are all micro-scalenes or backed bladelets 41 which would be compatible with these dates. A final Mesolithic element also seems represented more generally amongst the lithic assemblage from the entire site, with the presence of both asymmetric micro-tranchets and four-sided pieces. Date No of dates 8 millennium th 1 7th millennium 2 6th millennium 3 th 5 millennium 4 Table 6.12. Radiocarbon dates from Trench 19 Less than 2km from Blick Mead, an assemblage from a large pit known as the Coneybury Anomaly has been claimed as unique evidence for contact between Final Mesolithic and Neolithic groups. The faunal remains have been seen as the remains of a feast: the Neolithic groups providing cattle, Mesolithic groups roe deer (Gron et al. 2018). A Mesolithic input has been claimed on the basis that the lithics show Mesolithic traits and a wild fauna component is rare in early Neolithic assemblages; yet an assemblage focused on roe deer would also be unprecedented in local late Mesolithic contexts. The species is rare at Blick Mead, with only 8 examples out of 271 identified fragments. Further study of the lithic assemblage, which was not examined in this study and but is suggested to be, would resolve this issue. If this assemblage derives from a Mesolithic tradition it may reinforce hints of a fifth millennium interest in large pits and shafts in this region. A feature more securely associated with Mesolithic activity is the Fir Tree Field Shaft, a solution pit 5m wide. It has infilled with a chalk rubble, weathered from the sides, within which are several layers of soil, also slumped in from above, the lowest at 10m deep (Green and Allen 1998). The soil layers frequently contained charcoal and burnt flint, a common feature of Mesolithic pit deposits (Blinkhorn et al. 2016). Two roe deer skeletons, one at 7m deep, the other at 5.2m, have been dated to the final centuries of the Mesolithic (Griffiths 2011). These have been seen as natural deaths of animals that fell into the shaft and were trapped, however given the rarity of this species in Mesolithic faunal assemblages and the association of roe deer with the Coneybury anomaly, these could hint at practices of formal deposition. Further up the shaft, at the base of a weathering cone, a group of straight backed bladelets, elements of a possible composite object, were recovered. The date of this has been modelled to 4160-3980 cal BC (Griffiths 2011, 233). In the layer above the microliths, early Neolithic activity has been recorded. This took place 0-40 years (68% probability) after the microliths were left behind. Griffiths (2011, 296) has suggested the Fir Tree Field Shaft may also have been a place where groups with differing histories met in the years around 4000BC. Further west in Devon at Little Dartmouth Farm, the large 1.5m deep sixth millennium pit (see chapter 5) was recut for a final time in the early fifth millennium (Green et al. 2012). Moving further west, the little evidence that can be associated with the fifth millennium also suggests landscape continuity. The exception may be the shell midden sites where, apart from Blashenwell, there are no fifth millennium dates, though the presence of four-sided microliths at Culverwell may suggest some late occupation. In the case of Westward Ho! this is likely to be as a result of rapidly changing coastal geomorphology; by the fifth millennium peat had started to grow over the midden. Elsewhere there is more continuity: in addition to the recut of the Little Dartmouth Farm Pit, there is a late fifth millennium date from Birdcombe, in Somerset. Continuity is also seen in areas where the landscape was changing. On the Somerset Levels an estuarine phase of mudflats and saltmarshes between 5500 and 4500 cal BC was 42 followed by a slowing of sea-level rise and the development of extensive areas of reedswamp (Bell et al. 2015). Recent work by Bell and colleagues has recovered traces of post-transgression Mesolithic occupation: a lithic scatter in the base of a peat dated to the last couple of centuries of the Mesolithic at Chedzoy and a backed bladelet associated with a landsurface post-dating the transgression but sealed by an early Neolithic peat at Shapwick (Bell et al 2015). Plant macrofossils from late fifth millennium peat at Chedzoy are indicative of wetland fen, with lime-dominated deciduous woodland on the adjacent sandy island, while pollen from Shapwick indicates late Mesolithic damp woodland growing in fen carr with expanses of reeds and sedge swamp. The evidence suggests similar small-scale use of the area to the Late Mesolithic, focused on fishing and fowling. In Cornwall probable Final Mesolithic microliths have been recovered from the late Mesolithic site of Polowrian (Smith and Harris 1982, see chapter 5). A second site on the Lizard, Windmill Farm, has seven radiocarbon dates spanning the fifth millennium, as well as two dates showing evidence for earlier occupation at the start of the sixth millennium. The site located on flat ground by a stream, just before it dips into a boggy valley that leads down to the sea. Ploughing in 1982 uncovered lithic material that appeared to represent the edge of a scatter that continued to the west and north into an adjacent unploughed area. Following fieldwalking of the ploughzone, which suggested the presence of two separate scatters, a 10x11m trench was excavated by the Central Excavation Unit and the Cornwall Archaeological Society in the area immediately to the north. The excavations located a significant site with an assemblage of nearly 80,000 lithic artefacts and a rich assemblage of coarse stone tools (Smith 1984, Smith in press). A number of small shallow hollows contained fragments of oak charcoal, charred hazelnut shells and small pebbles c.10mm in diameter. These latter have been suggested perhaps to be gastroliths from the stomachs of seals (following Jones et al. 2019). Tasks at the site seem focused on microlith production. 1592 microburins and 968 microliths were recovered, despite only a sample of the excavated sediments being sieved. Beyond microliths, the denticulate scrapers, characteristic of the Mesolithic of the region, are also common with 246 recovered. Other tools such as scrapers and notches are present in smaller numbers. Bevel- ended and other coarse stone tools, with macrowear suggesting a variety of different functions, are well-represented. Unused examples are common, suggesting they may have been collected and cached for future use at the site. Microliths are mainly focused in the southwest of the main trench, surrounding two of the hollows. Denticulate scrapers and coarse stone tools are spread more evenly throughout the site (Smith in press). Proportions of tertiary bladelets are higher in the north of the site and microburins are also most common here, suggesting this is an area to which bladelet blanks were imported for microlith production. South Wales At Goldcliff intermittent occupation continued, though environments were changing. There are fifth millennium dates for site J, a site lacking earlier radiocarbon dates, though on stratigraphic grounds occupation here probably also extended back into the sixth millennium. Site J is on the edge of Goldcliff island and the highest of the sites investigated, at 1.5m asl (Bell 2007). It was also the most extensively investigated with 64m2 excavated. It lay in an area of hazel woodland, probably close to a spring. Charcoal indicates that oak, elm, ash, blackthorn, elder and ivy were also present nearby. The site was rich in lithics and animal bone, and five distinct clusters of material were located within the buried soil horizon. Of these, the central cluster B is the most substantial and displays a sharp fall-off in material which may be the result of a tent wall. This structure would be about 3m in diameter and was not associated with stake holes or any other structural evidence. As for the Star Carr hut, 43 larger material seems to have been pushed to the edges of the structure. Large quantities of lithics were present including 28 cores and core fragments, a microlith and 19 other tools – scrapers, notches, denticulates and retouched pieces. Microwear suggests butchery and craftwork with scraping of siliceous plants. Very little microdebitage was present. perhaps suggesting knapping stations were located elsewhere and scatter B represents an area of tool use and caching. Animal bone, often calcinated and mostly consisting of deer remains, is present within the putative structure as are a number of enigmatic wooden artefacts. A boar’s tusk was found on the edge of scatter B, which could represent a formal deposit (Bell 2007, 79). Other scatters are more diffuse. Microliths are associated with scatter A, to the west of the structure. Three of these are remarkably similar scalene triangles which may represent a composite tool; one of these has traces of butchery suggesting they may have been hafted as a knife. Scatter A is also associated with animal processing and cooking activities, indicated by faunal remains and clusters of burnt stones. The latter may have been obtained from the Usk or the Ebbw, a journey, probably by canoe, given their weight of 5-10kg. A probable tree throw was located in area A whose fill contained flint along with bones red deer, pig and aurochs, and may have been used as a midden. The different species representation between scatters A and B may suggest separate occupation episodes, though a cluster of animal bones in the northernmost scatter (E) is similar to area A. A small cluster of fish bones was associated with knapping debris in the eastern scatter C. The evidence gives the impression of repeated small-scale occupation. Two dates are available on wooden artefacts from site J, one on the edge of scatter B, the other from scatter C, suggesting one or more of these visits dates to between 4940-4710 cal BC, though it is likely that this is just one of several late Mesolithic visits to the site. In the late sixth and early fifth millennium, site J would have been on the margins of the estuary, whose silts cover the lower part of this site; tidal influences appear to have reached it around 4700 cal BC. Silt input is present in the upper layers of the buried soil and there are episodes of peat growth in the silts, indicating fluctuation between marine input and periods of stabilisation. Smaller quantities of lithics, including retouched pieces, some deer bones and a 1.16m long piece of worked oak which may represent a spear or a digging stick were found in the estuarine silts, indicative of continued activity during the fifth millennium in this environment for specific tasks, perhaps in the context of more substantial occupation of the adjacent dryland zone. For many centuries this site would have been at the margins of the fluctuating waters of the estuary, and as a place of dryland in close proximity to wetland environments, may have been favoured for short stays, some of which, seasonality evidence suggests, took place in spring or early summer. Around 4800 cal BC, estuarine sedimentation ceased at Goldcliff as the waters regressed. As surfaces stabilised, peat began to develop across the lower lying areas of the landscape, though episodes of marine inundation continued to occur within its lowest-lying levels and areas of saltmarsh persisted. Initially reeds grew on the peat surface, with areas of tall-herb fen developing, where sedges, hemp agrimony and nettles grew. Aquatics such as pondweed indicate pools. This wetland vegetation seems to have been regularly burnt, indicated by large quantities of charcoal. At site J on the drier ground, reedswamp was succeeded by alder and willow carr and finally at the end of the Mesolithic by mixed deciduous woodland (figure 6.22). The marine regression and peat development saw a marked reduction in human activity. While charcoal has been found at high levels in the reed peat, there is very little associated with alder carr and subsequent developments. A handful of flints, a couple of pieces of worked wood and some charcoal fragments are known from these later levels at site J and there is similar ephemeral evidence to the west of the island 44 for occupation at this time. However, there is none of the debris typical of the short-term campsites that characterise sites on the flanks of the island in the sixth and early fifth millennium; by the end of the Mesolithic the role this place played seems to have changed. <insert figure 6.22 here> Figure 6.22. Tree stump in the peat of the upper submgered forest. Copyright Martin Bell Elsewhere in Wales there is more evidence for continuity. In the uplands, at Waun Fignen Felen, the discovery of an asymmetric micro-tranchet indicates continued ephemeral use of the uplands during the Final Mesolithic, while similar examples are known from upland areas at Nant Melyn, west of Aberdare and Mynydd Blaenrhondda, just south of Craig-y-Llyn (Barton et al. 1991). On the upland plateau on the other side of the Rhondda valley and close to a stream is a remarkable carved oak post from Maerdy wind farm (figure 6.23). Radiocarbon dating of sapwood suggests this object belongs to the last two centuries of the fifth millennium. The timber measures 1.7m in length and 0.26m in width, with one side lost through degradation. The post seems to be rounded at one end and possibly tapered at the other. The intact edge shows a length of chevron patterning and above this a concentric diamond, which has been interpreted as a possible eye (the other side of the post where a second eye might have been located has been lost). While natural causes for this patterning have been explored, it has been determined more likely to be the product of human action. The post may have originally stood up as a marker, or have been deposited in the bog (Scott Jones 2018). If the concentric diamond does indeed represent an eye, this and the length of the post might suggest this represents a person. It can perhaps be compared to the Shirgir idol (Lillie et al 2005, Zhilin et al. 2018), an early Holocene wooden post from Russia, several metres long, with a carved head and geometric designs on the body. There may also be connections with other British Mesolithic contexts where wooden posts may have been erected, at Stonehenge, and Bryn Celli Ddu (see chapters 4 and 5). One difference from these earlier examples is the species involved, with a move to oak from earlier use of pine. If the life histories of the trees used to form these posts was significant, this may suggest a shift in the importance of particular tree species, as indeed we see the significant of particular animal species changing across the period. <insert figure 6.23 here> Figure 6.23. The Maerdy post. Copyright Richard Scott Jones Continuity between late and final Mesolithic is also evinced in the areas of dense activity in Pembrokeshire (see chapter 5). There is a noticeable presence of small numbers of asymmetric micro-tranchet types on sites that also have evidence for Late Mesolithic occupation (see chapter 5). A single microtranchet has been recovered at The Nab Head showing that this site with its long history continued to be visited (David 2007). Several examples have been found at Penpant, Solva, which lies on a south facing slope close to springs at the head of a small valley cut by a stream that runs into St Bride’s Bay (David 2020). Cwm Bach I, another site with micro-tranchets, is in a very similar environmental situation. Other examples are known from Pointzcastle where Mesolithic material clusters around springs in a narrow valley (David 2007). There are also examples from inland locations, such as Cutty Bridge on the Western Cleddau river (David and Painter 2014). Textbox 6.6: The last pig 45 The only dated evidence for the Final Mesolithic in Pembrokeshire comes from Lydstep. Here the remains of a pig, associated with narrow backed bladelets was located in intertidal peats by Leach following a storm in 1917 (Leach 1918). The peats seem to have formed in shallow lagoons behind barriers such as storm beaches (Murphy et al. 2014), and pollen evidence suggests the area where the pig was found was alder carr with areas of reed and sedge swamp towards dryland. The pig has been directly dated to 4345–3950 cal. BC (5300±100; OxA-1412). The relatively small size of the pig has been the basis of a suggestion that this was a domesticated animal rustled from the earliest Neolithic groups in the region (Lewis 1992). The pig is presumed to have escaped its hunters but died of blood loss (Jacobi 1980). Chatterton (2005) has by contrast suggested this may represent an episode of formal deposition, with the pig laid in a small pool, weighed down with a tree trunk. This may be worth considering given the lack of representation of pig amongst footprints recently discovered at Lydstep, as well as at more distant, but more thoroughly researched intertidal wetland areas such as Goldcliff. Charcoal and burnt flints had been discovered in the same layer by Leach, only c.23m from the pig, while in 2010 human and animal footprints were exposed and evidence of burning of the vegetation has also been noted in environmental profiles (Murphy et al. 2014). This perhaps suggests that the pig can be put in the context of a reasonable level of human activity in the wetland. This increases the possibility that its deposition is the product of human action, though whether it tells us about the nature of relations between Mesolithic and Neolithic groups remains to be seen. If this pig is the result of an act of formal deposition, this event, at the very end of the Mesolithic, is a curious echo of one right at the start: the deposition of a composite red deer at Star Carr (chapter 2). While much has changed over 5300 years, some concerns perhaps united the first and last Mesolithic people. Neolithic epilogue In many accounts of the transition, the Mesolithic only seems to exist as a foil for the sort of Neolithic that various archaeologists have imagined; an account in which the Mesolithic is reduced to a ‘reasonable paragraph’ (Achebe 1958). The paucity of radiocarbon dates and typochronologies has meant that an unchanging Mesolithic is presented, with a few key sites combined to produce a narrative of an unchanging way of life that lasted millennia. The current narrative of the transition, funnelled through recent aDNA research (Brace et al. 2019), falls into this familiar pattern. Yet the DNA analysis includes a single fifth millennium individual (from Oronsay). While this individual shows Western Hunter-Gatherer (WHG) ancestry only, sequencing of two ‘Neolithic’ individuals from western Scotland indicated a WHG ancestor within four generations. These were individuals from Raschoille Cave who were buried in a Mesolithic shell midden. It can be no coincidence that more complex processes are evident in the only region where there is coverage on a more appropriate temporal scale. Elsewhere in Britain nothing is known of the genetic histories of fifth millennium individuals and when groups with particular lineages reached Britain. Brace and colleagues (2019) noted greater WHG ancestry amongst Neolithic individuals from southeast England. This could reflect longer histories of continental connections in this region and it 46 may be that the genetic makeup of fourth millennium individuals sequenced reflects fifth millennium connections between Britain and the Continent (Lawrence et al. in press). The Mesolithic was hugely diverse, but above all it had a history. The early pioneers had different concerns from the people of the hazelnut-loving Mesolithic house horizon and the southwestern groups who camped by colourful waters. Rather than continuing to pursue a lifestyle that had lasted millennia, in the fifth millennium, times were continuing to change. Some people were caught up in a north-west European network of marine voyaging: continental trapezes and transverse arrowheads were reimagined in an insular context; T- antler axes, a continental tool, were made in Scotland; cereals may have been grown in the Pennine uplands and in places, ostentatious versions of familiar things - middens, pits and posts - were made. Only when this geographical diversity, and historical complexity is taken into consideration and the fact that there were many different Final Mesolithics acknowledged, can the significance of the changes at the start of Fourth Millennium be adequately understood. 47
US